Updated on 2024/04/16

写真a

 
HIRASAWA, Kotaro
 
Affiliation
Faculty of Science and Engineering
Job title
Professor Emeritus
Degree
Systems Engineering Approach to Elevator Group Supervisory Control ( Kyushu University )

Research Experience

  • 1992
    -
    2002

    Kyushu University   Faculty of Information Science and Electrical Engineering

  • 1992
    -
    2002

    Professor, Graduate School of Information

  •  
     
     

    Science and Electrical Engineering

Education Background

  •  
    -
    1964

    Kyushu University   School of Engineering  

  •  
    -
    1964

    Kyushu University   Faculty of Engineering   Electrical Engineering  

Committee Memberships

  • 2006
    -
    2008

    日本学術振興会特別研究員等審査会 専門委員 2006 - 2008

  • 2006
    -
    2008

    国際事業委員会 書面審査員 2006 - 2008

  • 2002
    -
    2003

    北九州市科学技術振興施策検討委員会 委員 2002 - 2003

  • 2001
    -
    2002

    計測自動制御学会  九州支部長

  • 2001
    -
    2002

    SICE Society of Instrument and Control Engineers  Head of Kyushu Branch

  • 2000
    -
    2002

    電気設備学会  九州支部長

  • 1994
    -
    2002

    福岡市専門委員 委員 1994 - 2002

  • 2001
    -
     

    計測自動制御学会  評議員

  • 2000
    -
     

    電気安全九州委員会 電気保安功労者表彰審査委員会委員長 2000 -

  • 1999
    -
    2000

    東部清掃工場建設検討委員会 委員 1999 - 2000

  • 1997
    -
    2000

    福岡都市圏海水淡水化施設検討委員会 委員 1997 - 2000

  • 1997
    -
    1998

    熱分解ガス化溶融システム技術検討委員会 委員 1997 - 1998

  • 1996
    -
    1998

    国際自動制御連盟学術会議委員会 委員 1996 - 1998

  • 1997
     
     

    福岡市新食肉市場整備提案審査委員会 委員 1997 - 1997

  • 1995
    -
    1996

    福岡市東部臨海地区総合エネルギーシステム調査委員会 委員 1995 - 1996

▼display all

Professional Memberships

  •  
     
     

    電気設備学会

  •  
     
     

    情報処理学会

  •  
     
     

    電気学会

  •  
     
     

    IEEE

  •  
     
     

    計測自動制御学会

  •  
     
     

    ACM

  •  
     
     

    SICE Society of Instrument and Control Engineers

  •  
     
     

    Association for Computing Machinery

▼display all

Research Areas

  • Control and system engineering / Kansei informatics / Soft computing

Research Interests

  • 複雑系

  • システム情報(知識)処理

  • 遺伝アルゴリズム

  • ニューラルネットワーク

  • Complex Systems

  • Knowlege Engineering

  • Genetic Algorithm

  • Neural Networks

▼display all

Awards

  • IEEE 福岡支部学生研究奨励賞

    2006  

  • SICE 九州 優秀論文賞

    2001  

  • 電気学会 優秀論文賞

    2001  

  • IEEE SMC 優秀論文賞

    1995  

  • 関東地方発明表彰

    1993  

  • IEEE SMC 優秀論文賞

    1987  

  • 関東地方発明表彰(発明奨励賞)

    1985  

  • オーム技術賞

    1978  

  • 全国発明表彰(朝日新聞発明賞)

    1976  

  • 関東地方発明表彰(支部長賞)

    1970  

  • オーム技術賞

    1968  

▼display all

 

Books and Other Publications

  • 確率一般化学習ネットワーク

    森北出版  2006

  • Multi-Branch Neural Networks and its Application to Stock Price Prediction

    Springer  2005

  • Method for Applying Neural Networks to Control of Nonlinear Systems

    Springer  2004

  • 確率一般化学習ネットワーク

    共立出版  2000

Works

  • エレベータ群管理システムの研究

    2003
    -
     

  • Elevator Group Supervisory Control Systems

    2003
    -
     

  • 並列化に適した最適化手法の研究

    2002
     
     

  • Optimization with Parallel Algorithm

    2002
     
     

  • 自律分散型ロボット制御システム

    1996
    -
    2002

  • Autonomous Robot Control System

    1996
    -
    2002

  • 遺伝的共生アルゴリズムの実用化に関する研究

    1999
    -
    2000

  • 一般化学習ネットワークによるシステムのモデル化とインテリジェント制御に関する研究

    1999
    -
    2000

  • Application of Genetic Symbiotic Algorithm

    1999
    -
    2000

  • Modelling and Intelligent Control Using Universal Learning Network

    1999
    -
    2000

  • 次世代プラント制御方式の研究

    1998
    -
    1999

  • Plan Control of Next Generation

    1998
    -
    1999

▼display all

Research Projects

  • Genetic Network Programming

    科学研究費補助金

    Project Year :

    1999
    -
    2005
     

  • Genetic Network Programming

    Grant-in-Aid for Scientific Research

    Project Year :

    1999
    -
    2005
     

  • 適応的ランダム探索法

    科学研究費補助金

    Project Year :

    1993
    -
    2005
     

  • 共生学習進化型マルチエージェントシステム

    科学研究費補助金

    Project Year :

    1993
    -
    2005
     

  • 一般化学習ネットワーク

    科学研究費補助金

    Project Year :

    1993
    -
    2005
     

  • Random Search with Intensification and Diversification

    Grant-in-Aid for Scientific Research

    Project Year :

    1993
    -
    2005
     

  • Multi-Agent Systems with Symbiotic Learning and Evolution

    Grant-in-Aid for Scientific Research

    Project Year :

    1993
    -
    2005
     

  • Universal Learning Network

    Grant-in-Aid for Scientific Research

    Project Year :

    1993
    -
    2005
     

▼display all

Misc

  • Efficient program generation by evolving graph structures with multi-start nodes

    Shingo Mabu, Kotaro Hirasawa

    APPLIED SOFT COMPUTING   11 ( 4 ) 3618 - 3624  2011.06

     View Summary

    Automatic program generation is one of the applicable fields of evolutionary computation, and Genetic Programming (GP) is the typical method for this field. On the other hand, Genetic Network Programming (GNP) has been proposed as an extended algorithm of GP in terms of gene structures. GNP is a graph-based evolutionary algorithm and applied to automatic program generation in this paper. GNP has directed graph structures which have some features inherently, for example, re-usability of nodes and the small number of nodes. These features contribute to creating complicated programs with compact structures and never cause bloat. In this paper, the extended algorithm of GNP is proposed, which can create plural programs simultaneously in one individual by using multi-start nodes. In addition, GNP can evolve the programs in one individual considering the fitness and also its standard deviation in order to evolve the plural programs efficiently. In the simulations, Even-n-Parity problem and Mirror Symmetry problem are used for the performance evaluation, and the results show that the proposed method outperforms the standard GNP with single start node. (C) 2011 Elsevier B.V. All rights reserved.

    DOI

  • Efficient program generation by evolving graph structures with multi-start nodes

    Shingo Mabu, Kotaro Hirasawa

    Applied Soft Computing Journal   11 ( 4 ) 3618 - 3624  2011.06

     View Summary

    Automatic program generation is one of the applicable fields of evolutionary computation, and Genetic Programming (GP) is the typical method for this field. On the other hand, Genetic Network Programming (GNP) has been proposed as an extended algorithm of GP in terms of gene structures. GNP is a graph-based evolutionary algorithm and applied to automatic program generation in this paper. GNP has directed graph structures which have some features inherently, for example, re-usability of nodes and the small number of nodes. These features contribute to creating complicated programs with compact structures and never cause bloat. In this paper, the extended algorithm of GNP is proposed, which can create plural programs simultaneously in one individual by using multi-start nodes. In addition, GNP can evolve the programs in one individual considering the fitness and also its standard deviation in order to evolve the plural programs efficiently. In the simulations, Even-n-Parity problem and Mirror Symmetry problem are used for the performance evaluation, and the results show that the proposed method outperforms the standard GNP with single start node. © 2011 Elsevier B.V.

    DOI

  • Dynamic Optimal Route Search Algorithm for Car Navigation Systems with Preferences by Dynamic Programming

    Manoj Kanta Mainali, Shingo Mabu, Shanqing Yu, Shinji Eto, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   6 ( 1 ) 14 - 22  2011.01

     View Summary

    Optimal route search to the destination is one of the most important functions of car navigation devices. The development of road traffic infrastructure has made it possible to receive real-time information of the traffic situation. Route search algorithms for car navigation devices make use of this information to avoid the traffic congestions. Such algorithms should find the new optimal route efficiently when the traffic situation changes. Usually, the minimum traveling time or distance is considered to define the optimal route. However, the minimum traveling time or distance is not always what the user is looking for. The user may prefer to travel on a certain route even at the cost of traveling time or distance. Car navigation devices should consider such preferences when finding the optimal route. In this paper, we propose a dynamic programming algorithm to find the optimal route considering that it should deal with the changes of the traffic situation and multiple criteria. The proposed method uses the information from the previous computation to find the new optimal route considering user preferences when the traveling time of the road section changes. The proposed method was applied to a real road network to find the optimal route. Results show that the proposed method can find the user-preferred optimal route. Simulation results also show better calculation time of the proposed method compared to the Dijkstra algorithm. (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • No Time limit and Time Limt Model of Multiple Round Dutch Auction Based on Genetic Network Programming

    S. Mau, D. Yu, Y. Chuan, K. Hirasawa

    JACIII   15 ( 1 ) 3 - 12  2011

  • An Intrusion-Detection Model Based on Fuzzy Class-Association-Rule Mining Using Genetic Network Programming

    Shingo Mabu, Ci Chen, Nannan Lu, Kaoru Shimada, Kotaro Hirasawa

    IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART C-APPLICATIONS AND REVIEWS   41 ( 1 ) 130 - 139  2011.01

     View Summary

    As the Internet services spread all over the world, many kinds and a large number of security threats are increasing. Therefore, intrusion detection systems, which can effectively detect intrusion accesses, have attracted attention. This paper describes a novel fuzzy class-association-rule mining method based on genetic network programming (GNP) for detecting network intrusions. GNP is an evolutionary optimization technique, which uses directed graph structures instead of strings in genetic algorithm or trees in genetic programming, which leads to enhancing the representation ability with compact programs derived from the reusability of nodes in a graph structure. By combining fuzzy set theory with GNP, the proposed method can deal with the mixed database that contains both discrete and continuous attributes and also extract many important class-association rules that contribute to enhancing detection ability. Therefore, the proposed method can be flexibly applied to both misuse and anomaly detection in network-intrusion-detection problems. Experimental results with KDD99Cup and DARPA98 databases from MIT Lincoln Laboratory show that the proposed method provides competitively high detection rates compared with other machine-learning techniques and GNP with crisp data mining.

    DOI CiNii

  • Multicar elevator group supervisory control system using genetic network programming

    Lu Yu, Shingo Mabu, Kotaro Hirasawa

    IEEJ Transactions on Electrical and Electronic Engineering   6 ( 1 ) S65 - S73  2011

     View Summary

    Elevator group control systems are the transportation systems for handling passengers in the buildings. With the increasing demand for high-rise buildings, it becomes important to improve the elevator service. The multicar elevators consist of plural cars in a single elevator shaft. It contributes to the improvement in passengers' handling capacity, while allowing the reduction of the space occupied in the building. In contrast with traditional elevator systems, the cars can no longer operate freely, where there are several restrictions on their available movements. This requires a more difficult control including stochastic scheduling with high combinatorial complexity in order to make the system more flexible. At present, lots of buildings with more than 40 floors are being built, which are usually divided into several zones served by local elevator groups. In addition, the cars should be operated at equal time intervals, especially in such a building with multicar elevator systems (MCES) in order to obtain its good performance. Genetic network programming (GNP), one of the evolutionary computations, can realize a rule-based MCES due to its directed graph structure of the individual, which makes the system more flexible. This paper discusses MCES using GNP in high-rise buildings. Also, the positions of elevators are considered to avoid the bunching phenomenon. The performance of MCES is studied and compared with single-deck elevator system (SDES) and double-deck elevator system (DDES). © 2010 Institute of Electrical Engineers of Japan. Published by John Wiley &amp
    Sons, Inc. © 2010 Institute of Electrical Engineers of Japan.

    DOI

  • An Efficient Preprocessing Method for Suboptimal Route Computation

    Feng Wen, Shingo Mabu, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   6 ( S1 ) S50 - S56  2011

     View Summary

    In this paper, an effective genetic-based clustering algorithm (GCA) is proposed to preprocess the road network into a multilevel network that can reduce the route computation time substantially. Based on the multilevel network, route computation algorithms can provide suboptimal routes with little loss of accuracy (LOA). The geographic information of the road network is considered when constructing the multilevel network. In the proposed GCA, three criteria are considered to evaluate the clustering results. Route calculation results based on the multilevel network constructed by the proposed GCA are compared with the recent research to analyze and quantify the LOA. (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI

  • An Intrusion-Detection Model Based on Fuzzy Class-AssociationRule Mining Using Genetic Network Programming

    S. Mabu, C. Chen, N. Lu, K. Hirasawa

    IEEE trans. System, Man and Cybernetics-Part C   41 ( 1 ) 130 - 139  2011

    DOI CiNii

  • A Genetic Algorithm Based Clustering Method for Optimal Route Calculation on Multilevel Networks

    WEN Feng, MABU Shingo, HIRASAWA Kotaro

    SICE JCMSI   4 ( 1 ) 83 - 88  2011

     View Summary

    This paper introduces a multilevel road network model to speed up the optimal route calculation on large size road networks by preprocessing and pre-computing. The multilevel road network is constructed by separating a large network into several smaller sub-networks and pushing up the boundary nodes. The new road sections on the multilevel network are decided by pre-computing the shortest path between the boundary nodes of each sub-network in a hierarchical manner. To optimize the structure of the multilevel road network, an effective genetic algorithm based clustering method is proposed, which can reduce the number of boundary nodes and road sections between boundary nodes of each sub-network greatly. Experimental results show that the proposed approach significantly reduces the search space of the route calculation algorithm over existing methods.

    DOI CiNii

  • Q value-based Dynamic Programming with Boltzmann Distribution in Large Scale Road Network

    YU S.

    SICE Journal of Control, Measurement, and System Integration   4 ( 2 ) 129 - 136  2011

    DOI CiNii

  • A novel evolutionary method to search interesting association rules by keywords

    G. Yang, S. Mabu, K. Shimada, K. Hirasawa

    Expert Systems with Applications   1 ( 1 ) 1 - 8  2011

    DOI

  • Dynamic Optimal Route Search Algorithm for Car Navigation Systems with Preferences by Dynamic Programming

    Manoj Kanta Mainali, Shingo Mabu, Shanqing Yu, Shinji Eto, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   6 ( 1 ) 14 - 22  2011.01

     View Summary

    Optimal route search to the destination is one of the most important functions of car navigation devices. The development of road traffic infrastructure has made it possible to receive real-time information of the traffic situation. Route search algorithms for car navigation devices make use of this information to avoid the traffic congestions. Such algorithms should find the new optimal route efficiently when the traffic situation changes. Usually, the minimum traveling time or distance is considered to define the optimal route. However, the minimum traveling time or distance is not always what the user is looking for. The user may prefer to travel on a certain route even at the cost of traveling time or distance. Car navigation devices should consider such preferences when finding the optimal route. In this paper, we propose a dynamic programming algorithm to find the optimal route considering that it should deal with the changes of the traffic situation and multiple criteria. The proposed method uses the information from the previous computation to find the new optimal route considering user preferences when the traveling time of the road section changes. The proposed method was applied to a real road network to find the optimal route. Results show that the proposed method can find the user-preferred optimal route. Simulation results also show better calculation time of the proposed method compared to the Dijkstra algorithm. (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • No Time limit and Time Limt Model of Multiple Round Dutch Auction Based on Genetic Network Programming

    S. Mau, D. Yu, Y. Chuan, K. Hirasawa

    JACIII   15 ( 1 ) 3 - 12  2011

  • An Intrusion-Detection Model Based on Fuzzy Class-Association-Rule Mining Using Genetic Network Programming

    Shingo Mabu, Ci Chen, Nannan Lu, Kaoru Shimada, Kotaro Hirasawa

    IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART C-APPLICATIONS AND REVIEWS   41 ( 1 ) 130 - 139  2011.01

     View Summary

    As the Internet services spread all over the world, many kinds and a large number of security threats are increasing. Therefore, intrusion detection systems, which can effectively detect intrusion accesses, have attracted attention. This paper describes a novel fuzzy class-association-rule mining method based on genetic network programming (GNP) for detecting network intrusions. GNP is an evolutionary optimization technique, which uses directed graph structures instead of strings in genetic algorithm or trees in genetic programming, which leads to enhancing the representation ability with compact programs derived from the reusability of nodes in a graph structure. By combining fuzzy set theory with GNP, the proposed method can deal with the mixed database that contains both discrete and continuous attributes and also extract many important class-association rules that contribute to enhancing detection ability. Therefore, the proposed method can be flexibly applied to both misuse and anomaly detection in network-intrusion-detection problems. Experimental results with KDD99Cup and DARPA98 databases from MIT Lincoln Laboratory show that the proposed method provides competitively high detection rates compared with other machine-learning techniques and GNP with crisp data mining.

    DOI CiNii

  • Multicar Elevator Group Supervisory Control System using Genetic Network Programming

    Lu Yu, Shingo Mabu, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   6 ( S1 ) S65 - S73  2011

     View Summary

    Elevator group control systems are the transportation systems for handling passengers in the buildings. With the increasing demand for high-rise buildings, it becomes important to improve the elevator service. The multicar elevators consist of plural cars in a single elevator shaft. It contributes to the improvement in passengers' handling capacity, while allowing the reduction of the space occupied in the building. In contrast with traditional elevator systems, the cars can no longer operate freely, where there are several restrictions on their available movements. This requires a more difficult control including stochastic scheduling with high combinatorial complexity in order to make the system more flexible. At present, lots of buildings with more than 40 floors are being built, which are usually divided into several zones served by local elevator groups. In addition, the cars should be operated at equal time intervals, especially in such a building with multicar elevator systems (MCES) in order to obtain its good performance. Genetic network programming (GNP), one of the evolutionary computations, can realize a rule-based MCES due to its directed graph structure of the individual, which makes the system more flexible. This paper discusses MCES using GNP in high-rise buildings. Also, the positions of elevators are considered to avoid the bunching phenomenon. The performance of MCES is studied and compared with single-deck elevator system (SDES) and double-deck elevator system (DDES). (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI

  • An Efficient Preprocessing Method for Suboptimal Route Computation

    Feng Wen, Shingo Mabu, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   6 ( S1 ) S50 - S56  2011

     View Summary

    In this paper, an effective genetic-based clustering algorithm (GCA) is proposed to preprocess the road network into a multilevel network that can reduce the route computation time substantially. Based on the multilevel network, route computation algorithms can provide suboptimal routes with little loss of accuracy (LOA). The geographic information of the road network is considered when constructing the multilevel network. In the proposed GCA, three criteria are considered to evaluate the clustering results. Route calculation results based on the multilevel network constructed by the proposed GCA are compared with the recent research to analyze and quantify the LOA. (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI

  • An Intrusion-Detection Model Based on Fuzzy Class-AssociationRule Mining Using Genetic Network Programming

    S. Mabu, C. Chen, N. Lu, K. Hirasawa

    IEEE trans. System, Man and Cybernetics-Part C   41 ( 1 ) 130 - 139  2011

    DOI CiNii

  • A genetic Algorithm Based Clusteruing Method for Optimal Route Calculation on Multilevel Networks

    F. Wen, S. Mabu, K. Hirasawa

    SICE JCMSI   4 ( 1 ) 83 - 88  2011

     View Summary

    This paper introduces a multilevel road network model to speed up the optimal route calculation on large size road networks by preprocessing and pre-computing. The multilevel road network is constructed by separating a large network into several smaller sub-networks and pushing up the boundary nodes. The new road sections on the multilevel network are decided by pre-computing the shortest path between the boundary nodes of each sub-network in a hierarchical manner. To optimize the structure of the multilevel road network, an effective genetic algorithm based clustering method is proposed, which can reduce the number of boundary nodes and road sections between boundary nodes of each sub-network greatly. Experimental results show that the proposed approach significantly reduces the search space of the route calculation algorithm over existing methods.

    DOI CiNii

  • Q Value-Based Dynamic Programming with Boltzmann Distribution in large Scale Road Network

    S. Yu, Y. Xu, S. Mabu, M. K. Mainali, K. Shimada, K. Hirasawa

    SICE JCMSI   4 ( 2 ) 129 - 136  2011

    DOI CiNii

  • A novel evolutionary method to search interesting association rules by keywords

    G. Yang, S. Mabu, K. Shimada, K. Hirasawa

    Expert Systems with Applications   1 ( 1 ) 1 - 8  2011

    DOI

  • A model of portfolio optimization using time adapting genetic network programming

    Yan Chen, Shingo Mabu, Kotaro Hirasawa

    COMPUTERS & OPERATIONS RESEARCH   37 ( 10 ) 1697 - 1707  2010.10

     View Summary

    This paper describes a decision-making model of dynamic portfolio optimization for adapting to the change of stock prices based on an evolutionary computation method named genetic network programming (GNP). The proposed model, making use of the information from technical indices and candlestick chart, is trained to generate portfolio investment advice. Experimental results on the Japanese stock market show that the decision-making model using time adapting genetic network programming (TA-GNP) method outperforms other traditional models in terms of both accuracy and efficiency. A comprehensive analysis of the results is provided, and it is clarified that the TA-GNP method is effective on the portfolio optimization problem. (C) 2009 Elsevier Ltd. All rights reserved.

    DOI

  • A model of portfolio optimization using time adapting genetic network programming

    Yan Chen, Shingo Mabu, Kotaro Hirasawa

    COMPUTERS & OPERATIONS RESEARCH   37 ( 10 ) 1697 - 1707  2010.10

     View Summary

    This paper describes a decision-making model of dynamic portfolio optimization for adapting to the change of stock prices based on an evolutionary computation method named genetic network programming (GNP). The proposed model, making use of the information from technical indices and candlestick chart, is trained to generate portfolio investment advice. Experimental results on the Japanese stock market show that the decision-making model using time adapting genetic network programming (TA-GNP) method outperforms other traditional models in terms of both accuracy and efficiency. A comprehensive analysis of the results is provided, and it is clarified that the TA-GNP method is effective on the portfolio optimization problem. (C) 2009 Elsevier Ltd. All rights reserved.

    DOI

  • Network Intrusion Detection Using Class Association Rule Mining Based on Genetic Network Programming

    Ci Chen, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   5 ( 5 ) 553 - 559  2010.09

     View Summary

    Because of the expansion of the Internet in recent years, computer systems are exposed to an increasing number and type of security threats. How to detect network intrusions effectively becomes an important technique. This paper proposes a class association rule mining approach based on genetic network programming (GNP) for detecting network intrusions. This approach can deal with both discrete and continuous attributes in network-related data. And it can be flexibly applied to both misuse detection and anomaly detection. Experimental results with KDD99Cup and DARPA98 database from MIT Lincoln Laboratory shows that the proposed method provides a competitive high detection rate (DR) compared to other machine learning techniques. (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • EvoCMAR: A New Evolutionary Method to Directly Mine Association Rules for Classification

    Guangfei Yang, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   5 ( 5 ) 574 - 585  2010.09

     View Summary

    In this paper, we propose an evolutionary method with a three-layer structure to directly mine association rules for classification. The association rules have been demonstrated to be useful for classification, such as classification based on association rule (CBA) and classification method based on multiple association rule (CMAR), and they are found to be more accurate than some traditional methods, such as C4.5. Generally speaking, there are two phases in an associative classification method: (i) association rules mining; (ii) classification by association rules. However, the two phases are almost separated, viz, during the first phase, the mining of association rules does not focus on classification. Moreover, when building the classifier in the second phase, most of the association rues will be pruned. As a result, if we are able to directly mine the classification association rules, we can save time. Meanwhile, we can expect even better accuracy because the mining procedure itself considers the classification. In this paper, we build a novel evolutionary method, named evolutionary classification method based on multiple association rule (EvoCMAR), to tackle these problems, and the simulation results show that it performs well in both accuracy and speed. (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • Network Intrusion Detection Using Class Association Rule Mining Based on Genetic Network Programming

    Ci Chen, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   5 ( 5 ) 553 - 559  2010.09

     View Summary

    Because of the expansion of the Internet in recent years, computer systems are exposed to an increasing number and type of security threats. How to detect network intrusions effectively becomes an important technique. This paper proposes a class association rule mining approach based on genetic network programming (GNP) for detecting network intrusions. This approach can deal with both discrete and continuous attributes in network-related data. And it can be flexibly applied to both misuse detection and anomaly detection. Experimental results with KDD99Cup and DARPA98 database from MIT Lincoln Laboratory shows that the proposed method provides a competitive high detection rate (DR) compared to other machine learning techniques. (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • EvoCMAR: A New Evolutionary Method to Directly Mine Association Rules for Classification

    Guangfei Yang, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   5 ( 5 ) 574 - 585  2010.09

     View Summary

    In this paper, we propose an evolutionary method with a three-layer structure to directly mine association rules for classification. The association rules have been demonstrated to be useful for classification, such as classification based on association rule (CBA) and classification method based on multiple association rule (CMAR), and they are found to be more accurate than some traditional methods, such as C4.5. Generally speaking, there are two phases in an associative classification method: (i) association rules mining; (ii) classification by association rules. However, the two phases are almost separated, viz, during the first phase, the mining of association rules does not focus on classification. Moreover, when building the classifier in the second phase, most of the association rues will be pruned. As a result, if we are able to directly mine the classification association rules, we can save time. Meanwhile, we can expect even better accuracy because the mining procedure itself considers the classification. In this paper, we build a novel evolutionary method, named evolutionary classification method based on multiple association rule (EvoCMAR), to tackle these problems, and the simulation results show that it performs well in both accuracy and speed. (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • Multiple-Round English Auction Agent Based on Genetic Network Programming

    Chuan Yue, Shingo Mabu, Yu Wang, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   5 ( 4 ) 450 - 458  2010.07

     View Summary

    The auction mechanism widely used in web-based sites, which is originally designed for human beings, might not be the most efficient one in the future, and there is a demand for evolutionary computation auction agents adaptable to the dynamic auction environments. in this paper, we have applied genetic network programming (GNP) to auction agents and developed multiple-round English auction mechanisms based on multi-agent systems. GNP is an evolutionary method that uses directed graph structures as genes to create compact optimal solutions by evolution. According to the simulation results, it has been found that the proposed method could help agents to evolve their strategies generation by generation to get more goods with less money. Also. GNP shows good performance in helping the agent to find out the most suitable strategy under the current situation. (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • Multiple-Round English Auction Agent Based on Genetic Network Programming

    Chuan Yue, Shingo Mabu, Yu Wang, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   5 ( 4 ) 450 - 458  2010.07

     View Summary

    The auction mechanism widely used in web-based sites, which is originally designed for human beings, might not be the most efficient one in the future, and there is a demand for evolutionary computation auction agents adaptable to the dynamic auction environments. in this paper, we have applied genetic network programming (GNP) to auction agents and developed multiple-round English auction mechanisms based on multi-agent systems. GNP is an evolutionary method that uses directed graph structures as genes to create compact optimal solutions by evolution. According to the simulation results, it has been found that the proposed method could help agents to evolve their strategies generation by generation to get more goods with less money. Also. GNP shows good performance in helping the agent to find out the most suitable strategy under the current situation. (C) 2010 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • Mining Fuzzy Association Rules: A General Model Based on Genetic Network Programming and its Applications

    Karla Taboada, Shingo Mabu, Eloy Gonzales, Kaoru Shimada, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   5 ( 3 ) 343 - 354  2010.05

     View Summary

    The initiative of combining association rule mining with fuzzy set theory has been applied frequently in recent years [1-5] The original idea conies horn cleating with quantitative attributes in a database, where discretization of the quantitative values into intervals would leach to (lintel or overestimation of the values that are neat the borders This is called the sharp boundary problem Fuzzy sets can help us to overcome this problem by allowing different degrees of the membership. not only I and 0 treated by traditional methods Attribute values can thereby be the membeis of mote than one set mid therefore give a more realistic view, on such data On the whet hand. fuzzy set theory has been shown to be a very useful tool in association rule mining. because the mined rules can be expressed in linguistic terms. winch ate more mutual and understandable lot human beings The linguistic representation is mainly useful when those discovered rules are presented to human experts for study In this impel a novel association rule mining approach that integrates the evolutionary optimization technique 'genetic network programming (GNP). and fuzzy set theory has been proposed lot mining interesting fuzzy rules from given quantitative dam The performance of our algorithm has been compared with other relevant algorithms and the experimental results show the advantages and effectiveness of the proposed model 2010 Institute of Electrical Engineers of Japan Published by John Wiley & Sons. Inc

    DOI CiNii

  • Mining Fuzzy Association Rules: A General Model Based on Genetic Network Programming and its Applications

    Karla Taboada, Shingo Mabu, Eloy Gonzales, Kaoru Shimada, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   5 ( 3 ) 343 - 354  2010.05

     View Summary

    The initiative of combining association rule mining with fuzzy set theory has been applied frequently in recent years [1-5] The original idea conies horn cleating with quantitative attributes in a database, where discretization of the quantitative values into intervals would leach to (lintel or overestimation of the values that are neat the borders This is called the sharp boundary problem Fuzzy sets can help us to overcome this problem by allowing different degrees of the membership. not only I and 0 treated by traditional methods Attribute values can thereby be the membeis of mote than one set mid therefore give a more realistic view, on such data On the whet hand. fuzzy set theory has been shown to be a very useful tool in association rule mining. because the mined rules can be expressed in linguistic terms. winch ate more mutual and understandable lot human beings The linguistic representation is mainly useful when those discovered rules are presented to human experts for study In this impel a novel association rule mining approach that integrates the evolutionary optimization technique 'genetic network programming (GNP). and fuzzy set theory has been proposed lot mining interesting fuzzy rules from given quantitative dam The performance of our algorithm has been compared with other relevant algorithms and the experimental results show the advantages and effectiveness of the proposed model 2010 Institute of Electrical Engineers of Japan Published by John Wiley & Sons. Inc

    DOI CiNii

  • A Bidding Strategy of Multiple Round Auctions based on Genetic Network Programming

    C. Yue, S. Mabu, D. Yu, Y. Wang, K. Hirasawa

    WCCI 2010   Barcelona   33 - 40  2010

    DOI

  • A Portfolio Selection Strategy Using Genetic Relation Algorithm

    Y. Chen, S. mabu, K. Hirasawa

    WCCI 2010   Barcelona  2010

    DOI

  • Classification Based on A Multiple-Dimentional Probability Distribution and Its Application to Network Intrusion Detection

    S. Mabu, W. Li, N. Nannan, Y. Wang, K. Hirasawa

    WCCI 2010   Barcelona   1437 - 1433  2010

    DOI

  • Generalized Rule Extraction and Traffic Prediction in the Optimal Route Serach

    H. Zhou, S. Mabu, X. Li, K. Shimada, K. Hirasawa

    WCCI 2010   Barcelona   2625 - 2632  2010

    DOI

  • Genetic network programming with estimation of distribution algorithms for class association rule mining in traffic prediction

    Xianneng Li, Shingo Mabu, Huiyu Zhou, Kaoru Shimada, Kotaro Hirasawa

    2010 IEEE World Congress on Computational Intelligence, WCCI 2010 - 2010 IEEE Congress on Evolutionary Computation, CEC 2010   Barcelona   2673 - 2680  2010

     View Summary

    As an extension of Genetic Algorithm (GA) and Genetic Programming (GP), a new approach named Genetic Network Programming (GNP) has been proposed in the evolutionary computation field. GNP uses multiple reusable nodes to construct directed-graph structures to represent its solutions. Recently, many research has clarified that GNP can work well in data mining area. In this paper, a novel evolutionary paradigm named GNP with Estimation of Distribution Algorithms (GNP-EDAs) is proposed and used to solve traffic prediction problems using class association rule mining. In GNP-EDAs, a probabilistic model is constructed by estimating the probability distribution from the selected elite individuals of the previous generation to replace the conventional genetic operators, such as crossover and mutation. The probabilistic model is capable of enhancing the evolution to achieve the ultimate objective. In this paper, two methods are proposed based on extracting the probabilistic information on the node connections and node transitions of GNP-EDAs to construct the probabilistic model. A comparative study of the proposed paradigm and the conventional GNP is made to solve the traffic prediction problems using class association rule mining. The simulation results showed that GNP-EDAs can extract the class association rules more effectively, when the number of the candidate class association rules increases. And the classification accuracy of the proposed method shows good results in traffic prediction systems. © 2010 IEEE.

    DOI

  • Genetic Network Programming with Generalized Rule Accumulation

    L. Wang, S. Mabu, Q. Meng, K. Hirasawa

    WCCI 2010   Barcelona   2681 - 2687  2010

    DOI

  • Guiding the Evolution of Genetic Network Programming with Reinforcement Learning

    Q. Meng, S. Mabu, Y. Wang, K. Hirasawa

    WCCI 2010   Barcelona   2778 - 2785  2010

    DOI

  • Multiple ODs Routing Algorithm for Traffic Systems using GA

    Y. Wang, S. Mabu, Q. Meng, M. K. Mainali, K. Hirasawa

    WCCI 2010   Barcelona   3425 - 3432  2010

    DOI

  • A method of association rule analysis for incomplete database using genetic network programming

    Kaoru Shimada, Kotaro Hirasawa

    Proceedings of the 12th Annual Genetic and Evolutionary Computation Conference, GECCO '10   Portland   1115 - 1122  2010

     View Summary

    A method of association rule mining from incomplete databases is proposed using Genetic Network Programming (GNP). GNP is one of the evolutionary optimization techniques, which uses the directed graph structure. An incomplete database includes missing data in some tuples. Previous rule mining approaches cannot handle incomplete data directly. The proposed method can extract rules directly from incomplete data without generating frequent itemsets used in conventional approaches. In this paper, the proposed method is combined with difference rule mining using GNP for flexible association analysis. We have evaluated the performances of the rule extraction from incomplete medical datasets generated by random missing values. In addition, artificial missing values for privacy hiding are considered using the proposed method. Copyright 2010 ACM.

    DOI

  • Pruning association rules using statistics and genetic relation algorithm

    Eloy Gonzales, Shingo Mabu, Karla Taboada, Kotaro Hirasawa, Kaoru Shimada

    Proceedings of the 12th Annual Genetic and Evolutionary Computation Conference, GECCO '10   Portland   419 - 420  2010

     View Summary

    Most of the classification methods proposed produces too many rules for humans to read over, that is, the number of generated rules is thousands or millions which means complex and hardly understandable for the users. In this paper, a new post-processing pruning method for class association rules is proposed by a combination of statistics and an evolutionary method named Genetic Relation Algorithm (GRA). The algorithm is carried out in two phases. In the first phase the rules are pruned depending on their matching degree and in the second phase GRA selects the most interesting rules using the distance between them and their strength.

    DOI

  • A Double-Deck Elevator Systems Controller with Idle cage Assignment Algorithm using Genetic Network Programming

    L. Yu, S. Mabu, J. Zhou, S. Eto, K. Hirasawa

    GECCO 2010   Portland   1313 - 1314  2010

    DOI

  • Parameters Tuning using RasID Algorithm in Q value-based Dynamic programming with Boltzmann Distribution

    S. Yu, S. Mabu, M. K. Mainali, K. Hirasawa

    SICE 2010   Taipei   162 - 167  2010

  • Generating Stock Trading Signals Baes on Matching Degree with Extracted Rules by Genetic Network Programming

    S. Mabu, Y. Lian, Y. Chen, K. Hirasawa

    SICE 2010   Taipei   1164 - 1169  2010

  • Web Mining using Genetic Relation Agorithm

    E. Gonzales, S. Mabu, K. Taboada, K. Hirasawa

    SICE 2010   Taipei   1622 - 1627  2010

  • Studies on Q Value-based Dynamic Programming with Boltzmann Distribution

    Y. Xu, D. Zhang, S. Mabu, S. Yu, K. Hirasawa, Y. Fang

    SICE 2010   Taipei   1628 - 1632  2010

  • Analysis of Various Interesting Measures in Classification Rule Mining for Traffic Prediction

    X. Li, S. Mabu, H. Zhou, K. Shimada, K. Hirasawa

    SICE 2010   Taipei   1969 - 1974  2010

  • Genetic Network Programming with Exception Control

    Q. meng, S. Mabu, K. Hirasawa

    SICE 2010   Taipei   2608 - 2613  2010

  • Hybrid Rule Mining based on Fuzzy GNP and Probabilistic Classification for Intrusion Detection

    N. Lu, S. Mabu, W. Li, K. Hirasawa

    SICE 2010   Taipei   2614 - 2619  2010

  • A Bidding Strategy based on Genetic Network Programming in Continuous Double Auctions

    C. Yue, S. Mabu, D. Yui, Y. Wang, K. Hirasawa

    SICE 2010   Taipei   2620 - 2625  2010

  • Functionally Distributed Syatems Using Parallel Genetic Network Programming

    Y. Zhang, X. Li, Y. Yang, S. Mabu, Y. Jin, K Hirasawa

    SICE 2010   Taipei   2626 - 2630  2010

  • Generalized Rule Accumulation Based On Genetic Network Programming Considering Different Population Size and Rule Length

    L. Wang, S. Mabu, F. Ye, K. Hirasawa

    SICE 2010   Taipei   2631 - 2636  2010

  • Time Related Association Rules Mining with Attribute Accumulation Mechanism Applied to Large Scale Traffic System

    X. Wang, S. Mabu, H. Zhou, K. Hirasawa

    SICE 2010   Taipei   2637 - 2641  2010

  • Face Recognition Using PCA with GNP Fuzzy Data Mining

    D. Zhang, S. Mabu, K. Taboada, F. Weng, K. Hirasawa

    SICE 2010   Taipei   3073 - 3077  2010

  • Enhancing Global Portfolio Optimization using Genetic Network Programming

    V. Parque, S. Mabu, K. Hirasawa

    SICE 2010   Taipei   3078 - 3083  2010

  • Trading Rules on Stock Markets Using Genetic Network Programming with Subroutines

    J. Li, Q. meng, Y. Yang, S. Mabu, Y. Wang, K. Hirasawa

    SICE 2010   Taipei   3084 - 3088  2010

  • Automatic Program Generation with Genetic Network Programming using Subroutines

    B. Li, S. Mabu, K. Hirasawa

    SICE 2010   Taipei   3089 - 3094  2010

  • Generating Trading Rules on the Stock Markets with Robust Genetic Network Progrtamming Using Variance of Fitness Values

    Y. Chen, K. Hirasawa

    SICE 2010   Taipei   3095 - 3102  2010

  • Evolutionary Approach for the Traffic Volume Estimation of Road Sections

    M. K. Mainali, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   100 - 105  2010

  • Optimal Route Planning with Restrictions for Car Navigation Systems

    M. K. Mainali, S. Mabu, X. Li, K Hirasawa

    IEEE SMC 2010   Estanbul   393 - 397  2010

  • Asset Selection in Global Financial Markets using Genetic Network Programming

    V. Parque, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   677 - 683  2010

  • GNP-Sarsa with Subroutines for Trading Rules on Stock Markets

    Y. Yang, J. Li, S. Mabu, K Hirasawa

    IEEE SMC 2010   Estanbul   1161 - 1165  2010

  • Multi-Car Elevator System using Genetic Network Programming for High-rise Building

    L. Yu, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   1216 - 1222  2010

  • Genetic Network Programming with Sarsa Learning Based Nonuniform Mutation

    Q. Meng, S. mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   1273 - 1278  2010

  • Evaluation on the Robustness of Genetic Network Programming with Reinforcement Learning

    S. Mabu, A. Tjahjadi, S. Sendari, K. Hirasawa

    IEEE SMC 2010   Estanbul   1659 - 1664  2010

    DOI

  • Double-Deck Elevator Systems with Idle Cage Assignment using Genetic Network Programming

    L. Yu, S. Mabu, J. Zhou, S. Eto, K Hirasawa

    IEEE SMC 2010   Estanbul   1987 - 1994  2010

  • Time Related Association Rule Mining with Accuracy Validation in Traffic Volume Predictionwith Large Scale Simulator

    H. Zhou, S. Mabu, K. Shimada, K. Hirasawa

    IEEE SMC 2010   Estanbul   2249 - 2255  2010

    DOI

  • Noise Reduction using Genetic Algorithm Based PCNN Method

    D. Zhang, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   2627 - 2633  2010

    DOI

  • Tile-World A Case Study of Genetic Network Programming with Automatic Program Generation

    B. Li, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   2708 - 2715  2010

  • Genetic Network Programming with New Genetic Operators

    F. Ye, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   3346 - 3353  2010

  • Various Temperature Parameter Control Methods in Q value-based Dynamic Programming with Boltzmann Distribution

    S. Yu, S. Mabu, M. K. Mainali, K. Shimada, K. Hirasawa

    IEEE SMC 2010   Estanbul   2677 - 2683  2010

    DOI

  • Robust genetic network programming on asset selection

    Victor Parque, Shingo Mabu, Kotaro Hirasawa

    IEEE Region 10 Annual International Conference, Proceedings/TENCON   Fukuoka   1021 - 1026  2010

     View Summary

    Financial innovation is continuously testing the asset selection models, which are the key both for building robust portfolios and for managing diversified risk. This paper describes a novel evolutionary based scheme for the asset selection using Robust Genetic Network Programming(r-GNP). The distinctive feature of r-GNP lies in its generalization ability when building the optimal asset selection model, in which several training environments are used throughout the evolutionary approach to avoid the over-fitting problem to the training data. Simulation using stocks, bonds and currencies in developed financial markets show competitive advantages over conventional asset selection schemes. © 2010 IEEE.

    DOI

  • Probabilistic Model Building Genetic Network Programming Using Multiple Probabilistic Vectors

    X. Li, S. Mabu, M. K. Mainali, K. Hirasawa

    TENCON 2010   Fukuoka   1398 - 1403  2010

    DOI

  • Genetic Network Programming with Route Nodes

    Fengming Ye, Shingo Mabu, Lutao Wang, Kotaro Hirasawa

    TENCON 2010: 2010 IEEE REGION 10 CONFERENCE   Fukuoka   1404 - 1409  2010

     View Summary

    Many classical methods such as Genetic Algorithm (GA), Genetic Programming (GP), Evolutionary Strategies (ES), etc. have accomplished significant contribution to the study of evolutionary computation. And in the past decade, a new approach named Genetic Network Programming (GNP) has been proposed. It is designed for especially solving complex problems in dynamic environments. Generally speaking, GNP is based on the algorithms of existed classical evolutionary computation techniques and uses the data structure of directed graphs which becomes the unique feature of GNP. So far, many studies have indicated that GNP can solve the complex problems in the dynamic environments very efficiently and effectively. Focusing on GNP's distinguished expression ability of the graph structure, this paper proposes an enhanced architecture for the standard GNP in order to improve the performance of GNP by using the exploited information extensively during the evolution process of GNP. In the enhanced architecture, the important gene information of the elite individuals is extracted and accumulated during evolution. And among the accumulated information, some of them are selected and encapsulated in the Route Nodes which are used to guide the evolution process. In this paper, the proposed architecture has been applied to the tile-world which is an excellent bench mark for evaluating the evolutionary computation architecture. The performance of the GNP with Route Nodes (GNP-RN) is compared with the conventional GNP. The simulation results show some merits of the proposed method over the conventional GNPs demonstrating its superiority.

    DOI

  • Variable Size Genetic Relation Algorithm for Portfolio Diversification

    V. Parque, S. Mabu, K. Hirasawa

    SCIS and ISIS 2010   Okayama   582 - 587  2010

  • Fuzzy Class Association Rule Mining for Traffic Prediction using Genetic Network Programming with Multi-Branch and Full-Path

    R. Nohmura, H. Zhou, S. Mabu, K. Hirasawa

    SCIS and ISIS 2010   Okayama   588 - 593  2010

  • Robustness Analysis of Genetic Network Programming with Reinforcement Learning

    A. Tjahjadi, S. Sendari, S. Mabu, K. Hirasawa

    SCIS and ISIS 2010   Fuluoka   594 - 601  2010

  • Multi-order Rule Extraction by Genetic Network Programming and Its Application to Stock Trading Problems

    Y. Xing, S. Mabu, L. Yuzhu, K. Hirasawa

    SCIS and ISIS 2010   Fukuoka   602 - 606  2010

  • Classification Based on the Distribution of Average Matching Degree and Its Application to Network Intrusion Detection

    T. Wang, M. Lu, S. Mabu, K. Hirasawa

    SCIS and ISIS 2010   Okayama   607 - 611  2010

  • Time related class association rule mining and its application to traffic prediction

    Huiyu Zhou, Shingo Mabu, Wei Wei, Kaoru Shimada, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   130 ( 2 ) 289 - 301  2010

     View Summary

    In this paper, an algorithm capable of finding important time related association rules is proposed where Genetic Network Programming (GNP) with Attribute Accumulation Mechanism (AAM) and Extraction Mechanism at Stages (EMS) is used. Then, the classification system based on extracted time related association rules is proposed to estimate to which class the current traffic data belong. Using this kind of classification mechanism, the traffic prediction is available since the rules extracted are based on time sequences. And, we also present the experimental results on the traffic prediction problem. © 2010 The Institute of Electrical Engineers of Japan.

    DOI CiNii

  • 重要度指標と調整ノードによる機能局在型 Genetic Network Programming

    間普真吾, 江藤慎治, 嶋田香, 平澤宏太郎

    信号処理   14 ( 1 ) 49 - 59  2010

  • Flexible Rule Mining for Difference Rules and Exception Rules from Incomplete Database

    SHIMADA Kaoru, HIRASAWA Kotaro

    IEEJ Transactions on Electronics, Information and Systems   130 ( 10 ) 1873 - 1881  2010

     View Summary

    Two flexible rule mining methods from incomplete database are proposed using Genetic Network Programing (GNP). GNP is one of the evolutionary optimization techniques, which uses the directed graph structure. One of the methods extracts the rules showing the different characteristics between different classes in a database. The method can obtain the rules like 'if P then Q' is interesting only in the focusing class. The other one mines interesting rules like even if itemset X and Y have weak or no statistical relation to class item C, the join of X and Y has strong relation to class item C. An incomplete database includes missing data in some tuples. Generally, it is not easy for Apriori-like methods to extract difference rules and exception rules from incomplete database. We have estimated the performances of the rule extraction using incomplete data in the environmental and medical field.

    DOI CiNii

  • Towards the Maintenance of Population Diversity: A Highbrid Probabilistic Model Building Genetic Network Programming

    X. Li, S. Mabu, K. Hirasawa

    進化計算学会学術論文誌   1 ( 1 ) 89 - 101  2010

  • Genetic Network Programming with Automatic Program Generation for Agent Control

    Li Bing, Mabu Shingo, Hirasawa Kotaro

    Transaction of the Japanese Society for Evolutionary Computation   1 ( 1 ) 43 - 53  2010

     View Summary

    In this paper, a new Genetic Network Programming with Automatic Program Generation (GNP-APG) has been proposed and applied to the Tileworld problem. A kind of genotype-phenotype mapping process is introduced in GNP-APG to create programs. The procedure of the program generation based on evolution is demonstrated in this paper. The advantages of the proposed method are also described. Simulations use different Tileworlds between the training phase and testing phase for performance evaluations and the results shows that GNP-APG could have better performances than the conventional GNP method.

    DOI CiNii

  • Intertransaction class association rule mining based on genetic network programming and its application to stock market prediction

    YANG Yuchen

    SICE JCMSI   3 ( 1 ) 50 - 58  2010

    DOI CiNii

  • An Evolutionary Negotiation Model using Genetic Network Programming

    T. Hossain, S. Mabu, K, Hirasawa

    Journal of Advanced Computaional Intelligence and Intelligent Informatics   14 ( 2 ) 215 - 223  2010

  • Genetic Network Programming with Reconstructed Individuals

    F. Ye, S. Mabu, L. Wang, S. Eto, K. Hirasawa

    SICE JCMSI   3 ( 2 ) 121 - 129  2010

    DOI

  • Evolving Asset Portfolios by Genetic Relation Algorithm

    V. Parque, S. Mabu, K Hirasawa

    JACIII   14 ( 5 ) 464 - 474  2010

    DOI

  • Genetic Network Programming with Estimation of Distribution Algorithms for Class Association Rule Mining in Traffic Prediction

    X. Li, S. Mabu, H. Zhou, K. Shimada, K Hirasawa

    JACIII   14 ( 5 ) 497 - 509  2010

  • A Double-Deck Elevator Systems Controller with Idle Cage Assignment Algorithm Using Genetic Network Programming

    S. Mabu, L. Yu, J. Zhou, S. Eto, K. Hirasawa

    JACIII   14 ( 5 ) 487 - 496  2010

  • Efficient Pruning of Class Asoociation Rules Using Statics and Genetic Relation Algorithm

    E. Gonzales, S. Mabu, K. Taboada, K. Shimada, K. Hirasawa

    SICE JCMSI   3 ( 5 ) 336 - 345  2010

    DOI CiNii

  • A Bidding Strategy of Multiple Round Auctions based on Genetic Network Programming

    C. Yue, S. Mabu, D. Yu, Y. Wang, K. Hirasawa

    WCCI 2010   Barcelona   33 - 40  2010

    DOI

  • A portfolio selection strategy using genetic relation algorithm

    Yan Chen, Shingo Mabu, Kotaro Hirasawa

    2010 IEEE World Congress on Computational Intelligence, WCCI 2010 - 2010 IEEE Congress on Evolutionary Computation, CEC 2010   Barcelona  2010

     View Summary

    This paper proposes a new strategy β-GRA for portfolio selection in which the return and risk are considered as measures of strength in Genetic Relation Algorithm (GRA). Since the portfolio beta β efficiently measures the volatility relative to the benchmark index or the capital market, β is usually employed for portfolio evaluation or prediction, but scarcely for portfolio construction process. The main objective of this paper is to propose an integrated portfolio selection strategy, which selects stocks based on β using GRA. GRA is a new evolutionary algorithm designed to solve the optimization problem due to its special structure. We illustrate the proposed strategy by experiments and compare the results with those derived from the traditional models. © 2010 IEEE.

    DOI

  • Classification Based on A Multiple-Dimentional Probability Distribution and Its Application to Network Intrusion Detection

    S. Mabu, W. Li, N. Nannan, Y. Wang, K. Hirasawa

    WCCI 2010   Barcelona   1437 - 1433  2010

    DOI

  • Generalized Rule Extraction and Traffic Prediction in the Optimal Route Serach

    H. Zhou, S. Mabu, X. Li, K. Shimada, K. Hirasawa

    WCCI 2010   Barcelona   2625 - 2632  2010

    DOI

  • Genetic network programming with estimation of distribution algorithms for class association rule mining in traffic prediction

    Xianneng Li, Shingo Mabu, Huiyu Zhou, Kaoru Shimada, Kotaro Hirasawa

    2010 IEEE World Congress on Computational Intelligence, WCCI 2010 - 2010 IEEE Congress on Evolutionary Computation, CEC 2010   Barcelona   2673 - 2680  2010

     View Summary

    As an extension of Genetic Algorithm (GA) and Genetic Programming (GP), a new approach named Genetic Network Programming (GNP) has been proposed in the evolutionary computation field. GNP uses multiple reusable nodes to construct directed-graph structures to represent its solutions. Recently, many research has clarified that GNP can work well in data mining area. In this paper, a novel evolutionary paradigm named GNP with Estimation of Distribution Algorithms (GNP-EDAs) is proposed and used to solve traffic prediction problems using class association rule mining. In GNP-EDAs, a probabilistic model is constructed by estimating the probability distribution from the selected elite individuals of the previous generation to replace the conventional genetic operators, such as crossover and mutation. The probabilistic model is capable of enhancing the evolution to achieve the ultimate objective. In this paper, two methods are proposed based on extracting the probabilistic information on the node connections and node transitions of GNP-EDAs to construct the probabilistic model. A comparative study of the proposed paradigm and the conventional GNP is made to solve the traffic prediction problems using class association rule mining. The simulation results showed that GNP-EDAs can extract the class association rules more effectively, when the number of the candidate class association rules increases. And the classification accuracy of the proposed method shows good results in traffic prediction systems. © 2010 IEEE.

    DOI

  • Genetic network programming with generalized rule accumulation

    Lutao Wang, Shingo Mabu, Qingbiao Meng, Kotaro Hirasawa

    2010 IEEE World Congress on Computational Intelligence, WCCI 2010 - 2010 IEEE Congress on Evolutionary Computation, CEC 2010   Barcelona   2681 - 2687  2010

     View Summary

    Genetic Network Programming(GNP) is a newly developed evolutionary computation method using a directed graph as its gene structure, which is its unique feature. It is competent for dealing with complex problems in dynamic environments and is now being well studied and applied to many real-world problems such as: elevator supervisory control, stock price prediction, traffic volume forecast and data mining, etc. This paper proposes a new method to accumulate evolutionary experiences and guide agent's actions by extracting and using generalized rules. Each generalized rule is a state-action chain which contains the past information and the current information. These generalized rules are accumulated and updated in the evolutionary period and stored in the rule pool which serves as an experience set for guiding new agent's actions. We designed a two-stage architecture for the proposed method and applied it to the Tile-world problem, which is an excellent benchmark for multi-agent systems. The simulation results demonstrated the efficiency and effectiveness of the proposed method in terms of both generalization ability and average fitness values and showed that the generalized rule accumulation method is especially remarkable when dealing with non-markov problems. © 2010 IEEE.

    DOI

  • Guiding the Evolution of Genetic Network Programming with Reinforcement Learning

    Q. Meng, S. Mabu, Y. Wang, K. Hirasawa

    WCCI 2010   Barcelona   2778 - 2785  2010

    DOI

  • Multiple ODs Routing Algorithm for Traffic Systems using GA

    Y. Wang, S. Mabu, Q. Meng, M. K. Mainali, K. Hirasawa

    WCCI 2010   Barcelona   3425 - 3432  2010

    DOI

  • A Method of Association Rule Analysis for Incomplete Database Using Genetic Network Programming

    K. Shimada, K. Hirasawa

    GECCO 2010   Portland   1115 - 1122  2010

    DOI

  • Pruning association rules using statistics and genetic relation algorithm

    Eloy Gonzales, Shingo Mabu, Karla Taboada, Kotaro Hirasawa, Kaoru Shimada

    Proceedings of the 12th Annual Genetic and Evolutionary Computation Conference, GECCO '10   Portland   419 - 420  2010

     View Summary

    Most of the classification methods proposed produces too many rules for humans to read over, that is, the number of generated rules is thousands or millions which means complex and hardly understandable for the users. In this paper, a new post-processing pruning method for class association rules is proposed by a combination of statistics and an evolutionary method named Genetic Relation Algorithm (GRA). The algorithm is carried out in two phases. In the first phase the rules are pruned depending on their matching degree and in the second phase GRA selects the most interesting rules using the distance between them and their strength.

    DOI

  • A double-deck elevator systems controller with idle cage assignment algorithm using genetic network programming

    Lu Yu, Shingo Mabu, Jin Zhou, Shinji Eto, Kotaro Hirasawa

    Proceedings of the 12th Annual Genetic and Evolutionary Computation Conference, GECCO '10   Portland   1313 - 1314  2010

     View Summary

    Many studies on Double-Deck Elevator Systems (DDES) have been done for exploring more efficient algorithms to improve the system transportation capacity, especially in a heavy traffic mode. The main idea of these algorithms is to decrease the number of stops during a round trip by grouping the passengers with the same destination as much as possible. How to dispatch idle cages, which is seldom considered in the heavy traffic mode, becomes important when developing the controller of DDES. In this paper, we propose a DDES controller with idle cage assignment algorithm using Genetic Network Programming (GNP) for a light traffic mode, which is based on a timer and event-driven hybrid model. To verify the efficiency and effectiveness of the proposed method, some experiments have been done under a special down-peak pattern.

    DOI

  • Parameters Tuning using RasID Algorithm in Q value-based Dynamic programming with Boltzmann Distribution

    S. Yu, S. Mabu, M. K. Mainali, K. Hirasawa

    SICE 2010   Taipei   162 - 167  2010

  • Generating Stock Trading Signals Baes on Matching Degree with Extracted Rules by Genetic Network Programming

    S. Mabu, Y. Lian, Y. Chen, K. Hirasawa

    SICE 2010   Taipei   1164 - 1169  2010

  • Web Mining using Genetic Relation Agorithm

    E. Gonzales, S. Mabu, K. Taboada, K. Hirasawa

    SICE 2010   Taipei   1622 - 1627  2010

  • Studies on Q Value-based Dynamic Programming with Boltzmann Distribution

    Y. Xu, D. Zhang, S. Mabu, S. Yu, K. Hirasawa, Y. Fang

    SICE 2010   Taipei   1628 - 1632  2010

  • Analysis of Various Interesting Measures in Classification Rule Mining for Traffic Prediction

    X. Li, S. Mabu, H. Zhou, K. Shimada, K. Hirasawa

    SICE 2010   Taipei   1969 - 1974  2010

  • Genetic Network Programming with Exception Control

    Q. meng, S. Mabu, K. Hirasawa

    SICE 2010   Taipei   2608 - 2613  2010

  • Hybrid Rule Mining based on Fuzzy GNP and Probabilistic Classification for Intrusion Detection

    N. Lu, S. Mabu, W. Li, K. Hirasawa

    SICE 2010   Taipei   2614 - 2619  2010

  • A Bidding Strategy based on Genetic Network Programming in Continuous Double Auctions

    C. Yue, S. Mabu, D. Yui, Y. Wang, K. Hirasawa

    SICE 2010   Taipei   2620 - 2625  2010

  • Functionally Distributed Syatems Using Parallel Genetic Network Programming

    Y. Zhang, X. Li, Y. Yang, S. Mabu, Y. Jin, K Hirasawa

    SICE 2010   Taipei   2626 - 2630  2010

  • Generalized Rule Accumulation Based On Genetic Network Programming Considering Different Population Size and Rule Length

    L. Wang, S. Mabu, F. Ye, K. Hirasawa

    SICE 2010   Taipei   2631 - 2636  2010

  • Time Related Association Rules Mining with Attribute Accumulation Mechanism Applied to Large Scale Traffic System

    X. Wang, S. Mabu, H. Zhou, K. Hirasawa

    SICE 2010   Taipei   2637 - 2641  2010

  • Face Recognition Using PCA with GNP Fuzzy Data Mining

    D. Zhang, S. Mabu, K. Taboada, F. Weng, K. Hirasawa

    SICE 2010   Taipei   3073 - 3077  2010

  • Enhancing Global Portfolio Optimization using Genetic Network Programming

    V. Parque, S. Mabu, K. Hirasawa

    SICE 2010   Taipei   3078 - 3083  2010

  • Trading Rules on Stock Markets Using Genetic Network Programming with Subroutines

    J. Li, Q. meng, Y. Yang, S. Mabu, Y. Wang, K. Hirasawa

    SICE 2010   Taipei   3084 - 3088  2010

  • Automatic Program Generation with Genetic Network Programming using Subroutines

    B. Li, S. Mabu, K. Hirasawa

    SICE 2010   Taipei   3089 - 3094  2010

  • Generating Trading Rules on the Stock Markets with Robust Genetic Network Progrtamming Using Variance of Fitness Values

    Y. Chen, K. Hirasawa

    SICE 2010   Taipei   3095 - 3102  2010

  • Evolutionary Approach for the Traffic Volume Estimation of Road Sections

    M. K. Mainali, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   100 - 105  2010

  • Optimal Route Planning with Restrictions for Car Navigation Systems

    M. K. Mainali, S. Mabu, X. Li, K Hirasawa

    IEEE SMC 2010   Estanbul   393 - 397  2010

  • Asset Selection in Global Financial Markets using Genetic Network Programming

    V. Parque, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   677 - 683  2010

  • GNP-Sarsa with Subroutines for Trading Rules on Stock Markets

    Y. Yang, J. Li, S. Mabu, K Hirasawa

    IEEE SMC 2010   Estanbul   1161 - 1165  2010

  • Genetic Network Programming with Sarsa Learning Based Nonuniform Mutation

    Q. Meng, S. mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   1273 - 1278  2010

  • Evaluation on the robustness of genetic network programming with reinforcement learning

    Shingo Mabu, Andre Tjahjadi, Siti Sendari, Kotaro Hirasawa

    Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics   Estanbul   1659 - 1664  2010

     View Summary

    Genetic Network Programming (GNP) has been proposed as one of the evolutionary algorithms and extended with reinforcement learning (GNP-RL). The combination of evolution and learning can efficiently evolve programs and the fitness improvement has been confirmed in the simulations of tileworld problems, elevator group supervisory control systems, stock trading models and wall following behavior of Kbepera robot. However, its robustness in testing environments has not been analyzed in detail yet. In this paper, the learning mechanism in the testing environment is introduced and it is confirmed that GNP-RL can show the robustness using a robot simulator WEBOTS, especially when unexperienced sensor troubles suddenly occur. The simulation results show that GNP-RL works well in the testing even if wrong sensor information is given because GNP-RL has a function to change programs using alternative actions automatically. In addition, the analysis on the effects of the parameters of GNP-RL is carried out in both training and testing simulations. ©2010 IEEE.

    DOI

  • Double-Deck Elevator Systems with Idle Cage Assignment using Genetic Network Programming

    L. Yu, S. Mabu, J. Zhou, S. Eto, K Hirasawa

    IEEE SMC 2010   Estanbul   1987 - 1994  2010

  • Time related association rule mining with accuracy validation in traffic volume prediction with large scale simulator

    Huiyu Zhou, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics   Estanbul   2249 - 2255  2010

     View Summary

    Genetic Network Programming(GNP) based time related association rules mining method provides an useful mean to investigate future traffic volumes of road networks and hence helps to develop traffic navigation systems. Further improvements have been proposed in this paper about the time related association rule mining using generalized GNP with Accuracy Validation. For better adapting to the real-time traffic situations of the large scale simulator, the mechanism of Accuracy Validation is studied. The aim of this algorithm is to better handle association rule extraction using prediction accuracy as criteria and guide the whole evolution process. The generalized algorithm which can find the important time related association rules is described and experimental results are presented considering a traffic prediction problem using the database provided by a large scale simulator SOUND/4U. ©2010 IEEE.

    DOI

  • Noise Reduction using Genetic Algorithm Based PCNN Method

    D. Zhang, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   2627 - 2633  2010

    DOI

  • Tile-World A Case Study of Genetic Network Programming with Automatic Program Generation

    B. Li, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   2708 - 2715  2010

  • Genetic Network Programming with New Genetic Operators

    F. Ye, S. Mabu, K. Hirasawa

    IEEE SMC 2010   Estanbul   3346 - 3353  2010

  • Various temperature parameter control methods in Q value-based dynamic programming with Boltzmann Distribution

    Shanqing Yu, Shingo Mabu, Manoj Kanta Mainali, Kaoru Shimada, Kotaro Hirasawa

    Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics   Estanbul   2677 - 2683  2010

     View Summary

    In order to alleviate the congestion in modern metropolises with over crowded traffics and improve the efficiency of Intelligent Transportation Systems, three temperature parameter control methods of Q value-based Dynamic Programming with Boltzmann Distribution have been proposed in this paper. T he simulation result shows that each method has its own areas of expertise depending on its features and all of the methods could improve the efficiency of the traffic system comparing with the conventional Greedy Method. ©2010 IEEE.

    DOI

  • Robust genetic network programming on asset selection

    Victor Parque, Shingo Mabu, Kotaro Hirasawa

    IEEE Region 10 Annual International Conference, Proceedings/TENCON   Fukuoka   1021 - 1026  2010

     View Summary

    Financial innovation is continuously testing the asset selection models, which are the key both for building robust portfolios and for managing diversified risk. This paper describes a novel evolutionary based scheme for the asset selection using Robust Genetic Network Programming(r-GNP). The distinctive feature of r-GNP lies in its generalization ability when building the optimal asset selection model, in which several training environments are used throughout the evolutionary approach to avoid the over-fitting problem to the training data. Simulation using stocks, bonds and currencies in developed financial markets show competitive advantages over conventional asset selection schemes. © 2010 IEEE.

    DOI

  • Genetic network programming with route nodes

    Fengming Ye, Shingo Mabu, Lutao Wang, Kotaro Hirasawa

    IEEE Region 10 Annual International Conference, Proceedings/TENCON   Fukuoka   1404 - 1409  2010

     View Summary

    Many classical methods such as Genetic Algorithm (GA), Genetic Programming (GP), Evolutionary Strategies (ES), etc. have accomplished significant contribution to the study of evolutionary computation. And in the past decade, a new approach named Genetic Network Programming (GNP) has been proposed. It is designed for especially solving complex problems in dynamic environments. Generally speaking, GNP is based on the algorithms of existed classical evolutionary computation techniques and uses the data structure of directed graphs which becomes the unique feature of GNP. So far, many studies have indicated that GNP can solve the complex problems in the dynamic environments very efficiently and effectively. Focusing on GNP's distinguished expression ability of the graph structure, this paper proposes an enhanced architecture for the standard GNP in order to improve the performance of GNP by using the exploited information extensively during the evolution process of GNP. In the enhanced architecture, the important gene information of the elite individuals is extracted and accumulated during evolution. And among the accumulated information, some of them are selected and encapsulated in the Route Nodes which are used to guide the evolution process. In this paper, the proposed architecture has been applied to the tile-world which is an excellent bench mark for evaluating the evolutionary computation architecture. The performance of the GNP with Route Nodes (GNP-RN) is compared with the conventional GNP. The simulation results show some merits of the proposed method over the conventional GNPs demonstrating its superiority. ©2010 IEEE.

    DOI

  • Variable Size Genetic Relation Algorithm for Portfolio Diversification

    V. Parque, S. Mabu, K. Hirasawa

    SCIS and ISIS 2010   Okayama   582 - 587  2010

  • Fuzzy Class Association Rule Mining for Traffic Prediction using Genetic Network Programming with Multi-Branch and Full-Path

    R. Nohmura, H. Zhou, S. Mabu, K. Hirasawa

    SCIS and ISIS 2010   Okayama   588 - 593  2010

  • Robustness Analysis of Genetic Network Programming with Reinforcement Learning

    A. Tjahjadi, S. Sendari, S. Mabu, K. Hirasawa

    SCIS and ISIS 2010   Fuluoka   594 - 601  2010

  • Multi-order Rule Extraction by Genetic Network Programming and Its Application to Stock Trading Problems

    Y. Xing, S. Mabu, L. Yuzhu, K. Hirasawa

    SCIS and ISIS 2010   Fukuoka   602 - 606  2010

  • Classification Based on the Distribution of Average Matching Degree and Its Application to Network Intrusion Detection

    T. Wang, M. Lu, S. Mabu, K. Hirasawa

    SCIS and ISIS 2010   Okayama   607 - 611  2010

  • Time related class association rule mining and its application to traffic prediction

    Huiyu Zhou, Shingo Mabu, Wei Wei, Kaoru Shimada, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   130 ( 2 ) 289 - 301  2010

     View Summary

    In this paper, an algorithm capable of finding important time related association rules is proposed where Genetic Network Programming (GNP) with Attribute Accumulation Mechanism (AAM) and Extraction Mechanism at Stages (EMS) is used. Then, the classification system based on extracted time related association rules is proposed to estimate to which class the current traffic data belong. Using this kind of classification mechanism, the traffic prediction is available since the rules extracted are based on time sequences. And, we also present the experimental results on the traffic prediction problem. © 2010 The Institute of Electrical Engineers of Japan.

    DOI CiNii

  • Towards the Maintenance of Population Diversity: A Highbrid Probabilistic Model Building Genetic Network Programming

    X. Li, S. Mabu, K. Hirasawa

    進化計算学会学術論文誌   1 ( 1 ) 89 - 101  2010

  • Genetic Network Programming with Automatic Program Generation for Agent Control

    B. Li, S. Mabu, K. Hirasawa

    進化計算学会学術論文誌   1 ( 1 ) 43 - 53  2010

  • Intertransaction Class Association Rule Mining Based on Genetic Network Programming and its Application to Stock Market Prediction

    Y. Yang, S. Mabu, K. Shimada, K Hirasawa

    SICE Journal of Control, Measurement, and System Integration   3 ( 1 ) 50 - 58  2010

    DOI CiNii

  • An Evolutionary Negotiation Model using Genetic Network Programming

    T. Hossain, S. Mabu, K, Hirasawa

    Journal of Advanced Computaional Intelligence and Intelligent Informatics   14 ( 2 ) 215 - 223  2010

  • Genetic Network Programming with Reconstructed Individuals

    F. Ye, S. Mabu, L. Wang, S. Eto, K. Hirasawa

    SICE JCMSI   3 ( 2 ) 121 - 129  2010

    DOI

  • Evolving asset portfolios by genetic relation algorithm

    Victor Parque, Shingo Mabu, Kotaro Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   14 ( 5 ) 464 - 474  2010

     View Summary

    Global financial development have opened innumerable risks and opportunities for investments. A global view of the portfolio allocation through diversification brings advantages for the risk allocation in investments. In this paper, an asset allocation framework under the return, risk and liquidity considerations is proposed for short term investment using Genetic Relation Algorithm. Simulations using the stocks, bonds and currencies from relevant financial markets in USA, Europe and Asia show that the proposed framework is effective and robust. The efficacy of the proposed method is compared against the relevant constructs in finance and computational fields.

    DOI

  • Genetic Network Programming with Estimation of Distribution Algorithms for Class Association Rule Mining in Traffic Prediction

    X. Li, S. Mabu, H. Zhou, K. Shimada, K Hirasawa

    JACIII   14 ( 5 ) 497 - 509  2010

  • A Double-Deck Elevator Systems Controller with Idle Cage Assignment Algorithm Using Genetic Network Programming

    S. Mabu, L. Yu, J. Zhou, S. Eto, K. Hirasawa

    JACIII   14 ( 5 ) 487 - 496  2010

  • Efficient Pruning of Class Asoociation Rules Using Statics and Genetic Relation Algorithm

    E. Gonzales, S. Mabu, K. Taboada, K. Shimada, K. Hirasawa

    SICE JCMSI   3 ( 5 ) 336 - 345  2010

    DOI CiNii

  • A portfolio optimization model using Genetic Network Programming with control nodes

    Yan Chen, Etsushi Ohkawa, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    EXPERT SYSTEMS WITH APPLICATIONS   36 ( 7 ) 10735 - 10745  2009.09

     View Summary

    Many evolutionary computation methods applied to the financial field have been reported. A new evolutionary method named "Genetic Network Programming" (GNP) has been developed and applied to the stock market recently. The efficient trading rules created by GNP has been confirmed in our previous research. In this paper a multi-brands portfolio optimization model based on Genetic Network Programming with control nodes is presented. This method makes use of the information from technical indices and candlestick chart. The proposed optimization model, consisting of technical analysis rules, are trained to generate trading advice. The experimental results on the Japanese stock market show that the proposed optimization system using GNP with control nodes method outperforms other traditional models in terms of both accuracy and efficiency. We also compared the experimental results of the proposed model with the conventional GNP based methods, GA and Buy&Hold method to confirm its effectiveness, and it is clarified that the proposed trading model can obtain much higher profits than these methods. (C) 2009 Elsevier Ltd. All rights reserved.

    DOI

  • A portfolio optimization model using Genetic Network Programming with control nodes

    Yan Chen, Etsushi Ohkawa, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    EXPERT SYSTEMS WITH APPLICATIONS   36 ( 7 ) 10735 - 10745  2009.09

     View Summary

    Many evolutionary computation methods applied to the financial field have been reported. A new evolutionary method named "Genetic Network Programming" (GNP) has been developed and applied to the stock market recently. The efficient trading rules created by GNP has been confirmed in our previous research. In this paper a multi-brands portfolio optimization model based on Genetic Network Programming with control nodes is presented. This method makes use of the information from technical indices and candlestick chart. The proposed optimization model, consisting of technical analysis rules, are trained to generate trading advice. The experimental results on the Japanese stock market show that the proposed optimization system using GNP with control nodes method outperforms other traditional models in terms of both accuracy and efficiency. We also compared the experimental results of the proposed model with the conventional GNP based methods, GA and Buy&Hold method to confirm its effectiveness, and it is clarified that the proposed trading model can obtain much higher profits than these methods. (C) 2009 Elsevier Ltd. All rights reserved.

    DOI

  • Stock Price Prediction using Neural Networks with RasID-GA

    Shingo Mabu, Yan Chen, Dongkyu Sohn, Kaoru Shimada, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   4 ( 3 ) 392 - 403  2009.05

     View Summary

    In general, neural networks are widely used in pattern recognition, system modeling and prediction, and can model complex nonlinear systems. In the previous work, we proposed a novel training algorithm, Adaptive Random Search with Intensification and Diversification combined with Genetic Algorithm (RasID-GA), for training the multibranch recurrent neural networks recently developed. In this paper, RasID-GA has been applied to predict stock market prices using the multibranch feed forward neural networks. We predicted the next day's closing stock price with several past closing stock prices. We used the stock prices of 20 brands for 720 days in order to evaluate the generalization ability of the proposed method. (C) 2009 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • Stock Price Prediction using Neural Networks with RasID-GA

    Shingo Mabu, Yan Chen, Dongkyu Sohn, Kaoru Shimada, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   4 ( 3 ) 392 - 403  2009.05

     View Summary

    In general, neural networks are widely used in pattern recognition, system modeling and prediction, and can model complex nonlinear systems. In the previous work, we proposed a novel training algorithm, Adaptive Random Search with Intensification and Diversification combined with Genetic Algorithm (RasID-GA), for training the multibranch recurrent neural networks recently developed. In this paper, RasID-GA has been applied to predict stock market prices using the multibranch feed forward neural networks. We predicted the next day's closing stock price with several past closing stock prices. We used the stock prices of 20 brands for 720 days in order to evaluate the generalization ability of the proposed method. (C) 2009 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • A Nonlinear Model to Rank Association Rules Based on Semantic Similarity and Genetic Network Programing

    Guangfei Yang, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   4 ( 2 ) 248 - 256  2009.03

     View Summary

    Many methods have been studied for mining association rules efficiently. However, because these methods usually generate a large number of rules, it is still a heavy burden for the users to find the most interesting ones. In this paper, we propose a novel method for finding what the user is interested in by assigning several keywords, like searching documents on the Web using search engines. By considering both the semantic similarity between the rules and keywords, and the statistical information like support, confidence, chi-squared value, etc. we could rank the rules by a new method named RuleRank, where evolutionary methods are applied to find the optimal ranking model. Experiments show that our approach is effective for the users to find what they want. (C) 2009 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons. Inc.

    DOI CiNii

  • エレベータの省エネ運転による省エネ効果と利用者の便益の変化に関する研究

    上野剛, 中野幸夫, 平澤宏太郎, L. Yu

    第28回エネルギー・資源学会研究発表会    2009

  • エレベータの省エネ運転による省エネ効果と利用者便益変化

    上野剛, 中野幸夫, 平澤宏太郎, L. Yu

    平成21年度空気調和・衛生工学会大会    2009

  • Genetic Network Programming for Fuzzy Association Rule-Based Classification

    K. Taboada, E. Gonzales, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2009   Norway   2387 - 2394  2009

    DOI

  • Mining Multi-Class Datasets using Genetic Relation Algorithm for Rule Reduction

    E. Gonzales, K. Taboada, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2009   Norway   3249 - 3255  2009

    DOI

  • Multi-Car Elevator Group Supervisory Control System using Genetic Network Programming

    L. Yu, S. Mabu, T. Zhang, S. Eto, K. Hirasawa

    CEC 2009   Norway   2188 - 2193  2009

    DOI

  • Genetic Network Programming with Reconstructed Individuals

    F. Ye, S. Mabu, L. Wang, S. Eto, K. Hirasawa

    CEC 2009   Norway   854 - 859  2009

    DOI

  • Generalized Time Related Sequential Association Rule Mining and Traffic Prediction

    H. Zhou, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2009   Norway   2654 - 2661  2009

    DOI

  • Constructing Portfolio Investment Strategy Based on Time Adaptive Genetic Network Programming

    Y. Chen, S. Mabu, E. Ohkawa, K. Hirasawa

    CEC 2009   Norway   2379 - 2386  2009

    DOI

  • Genetic Network Programming with Rule Accumulation Considering Judgment Order

    L. Wang, F. Ye, S. Mabu, K. Hirasawa

    CEC 2009   Norway   3176 - 3182  2009

    DOI

  • Generalized Association Rules Mining with Multi-Branches Full-Paths and Its Application to Traffic Volume Prediction

    Huiyu Zhou, Shingo Mabu, Manoj Kanta Mainali, Xianneng Li, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   147 - 152  2009

  • Traveling Time Prediction using Isolation Rules

    Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009     1523 - 1528  2009

  • Robust Genetic Network Programming Using SARSA Learning for Autonomous Robots

    Sung Gil Park, Shingo Mabu, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   523 - 527  2009

  • Rule Accumulation Method with Modified Fitness Function based on Genetic Network Programming

    Lutao Wang, Shingo Mabu, Fengming Ye, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   1000 - 1005  2009

  • Temperature Parameter Control of Q value-based Dynamic Programming with Boltzmann Distribution

    Shanqing Yu, Shingo Mabu, Manoj Kanta Mainali, Shinji Eto, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   1471 - 1476  2009

  • Stock Movement Prediction using Fuzzy Intertransaction Class Association Rule Mining based on Genetic Network Programming

    Yuchen Yang, Shingo Mabu, Etsushi Ohkawa, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   2561 - 2566  2009

  • Global Portfolio Diversification by Genetic Relation Algorithm

    Victor Parque, Shingo Mabu, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   2567 - 2572  2009

  • A Genetic Relation Algorithm with Guided Mutation for the Large-Scale Portfolio Optimization

    Yan Chen, Chuan Yue, Shingo Mabu, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   2579 - 2584  2009

  • A New Associative Classification Method by Integrating CMAR and An Evolutionary Three-layers Structure

    Guangfei Yang, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   2920 - 2925  2009

  • Genetic Network Programming with Estimation of Distribution Algorithms, and its Application to Association Rule Mining for Traffic Prediction

    Xianneng Li, Shingo Mabu, Huiyu Zhou, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009     3457 - 3462  2009

  • Intrusion Detection System Combining Misuse Detection and Anomaly Detection Using Genetic Network Programming

    Yunlu Gong, Shingo Mabu, Ci Chen, Yifei Wang, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3463 - 3467  2009

  • Time Related Association Rules Mining for Traffic Prediction based on Genetic Network Programming combined with Estimation of Distribution Algorithms

    Yang Wang, Shingo Mabu, Huiyu Zhou, Xianneng Li, Kaoru Shimada

    ICROS-SICE International Joint Conference 2009   Fukuoka   3468 - 3473  2009

  • Genetic Network Programming with General Individual Reconstruction

    Fengming Ye, Shingo Mabu, Lutao Wang, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3474 - 3479  2009

  • Analysis of Fuzzy Class Association Rule Mining Based on Genetic Network Programming

    Ci Chen, Shingo Mabu, Chuan Yue, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3480 - 3484  2009

  • Global Optimal Routing for Traffic Systems with Multiple ODs using Genetic Algorithm

    Yu Wang, Shingo Mabu, Chuan Yue, Manoj Kanta Mainali, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3731 - 3737  2009

  • Q value-based Dynamic Programming with Evolving Penalties for Road Networks

    Manoj Kanta Mainali, Shingo Mabu, Yu Wang, Shanqing Yu, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3738 - 3743  2009

  • Class Association Rule Mining with Correlation Measures using Genetic Network Programming

    Eloy Gonzales, Shingo Mabu, Karla Taboada, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3850 - 3856  2009

  • Agent Bidding Strategy of Multiple Round English Auction based on Genetic Network Programming

    Chuan Yue, Shingo Mabu, Yan Chen, Yu Wang, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3857 - 3862  2009

  • Fuzzy Association Rule Mining and Classifier with Chi-squared Correlation Measure using Genetic Network Programming

    Karla Taboada, Shingo Mabu, Eloy Gonzales, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3863 - 3869  2009

  • Adaptive Controller for Double-Deck Elevator System using Genetic Network Programming

    Johanna Mansilla, Shingo Mabu, Lu Yu, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3870 - 3873  2009

  • A New Associative Classification Method by Integrating CMAR and RuleRank Model based on Genetic Network Programming

    Guangfei Yang, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3874 - 3879  2009

  • Network intrusion detection using fuzzy class association rule mining based on genetic network programming

    Ci Chen, Shingo Mabu, Chuan Yue, Kaoru Shimada, Kotaro Hirasawa

    Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics   San Antonio Texas   60 - 67  2009

     View Summary

    Computer Systems are exposed to an increasing number and type of security threats due to the expanding of internet in recent years. How to detect network intrusions effectively becomes an important techniques. This paper presents a novel fuzzy class association rule mining method based on Genetic Network Programming(GNP) for detecting network intrusions. GNP is an evolutionary optimization techniques, which uses directed graph structures as genes instead of strings(Genetic Algorithm) or trees(Genetic Programming), leading to creating compact programs and implicitly memorizing past action sequences. By combining fuzzy set theory with GNP, the proposed method can deal with the mixed database which contains both discrete and continuous attributes. And it can be flexibly applied to both misuse and anomaly detection in Network Intrusion Detection Problem. Experimental results with KDD99Cup and DAPRA98 databases from MIT Lincoln Laboratory show that the proposed method provides a competitively high detection rate compared with other machine learning techniques. ©2009 IEEE.

    DOI

  • A Study on Energy Consumption of Elevator Group Supervisory Control Systems using Genetic Network Programming

    L. Yu, S. Mabu, T. Thang, K Hirasawa, T Ueno

    IEEE SMC 2009   San Antonio Texas  2009

    DOI

  • Backward Time Related Association Rule Mining with Database Arrangement in Traffic Volume Prediction

    H. Zhou, S. Mabu, K. Shimada, K. Hirasawa

    IEEE SMC 2009   San Antonio Texas  2009

    DOI

  • Evolving Plural Programs by Genetic Network Programming with Multi-Start Nodes

    S. Mabu, K. Hirasawa

    IEEE SMC 2009   San Antonio Texas  2009

    DOI

  • Fuzzy Classification Rule Mining Based on Genetic Network Programming Algorithm

    K. Toboada, S. Mabu, E. Gonzales, K. Shimada, K. Hirasawa

    IEEE SMC 2009   San Antonio Texas  2009

    DOI

  • Multi-Route Algorithm using Temperature Control of Boltzmann Distribution in Q Value-based Dynamic Programming

    S. Yu, S. mabu, M. K. Mainali, S. Eto, K. Shimada, K. Hirasawa

    IEEE SMC 2009   San Antonio Texas  2009

    DOI

  • A Portfolio Selection Model using Genetic Relation Algorithm and Genetic Network Programming

    Y. Chen, S. Mabu, K. Hirasawa

    IEEE SMC 2009   San Antonio Texas  2009

    DOI

  • Real time Updating Genetic Network Programming for adapting to the change of stock prices

    Yan Chen, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   129 ( 2 ) 22 - 354  2009

     View Summary

    The key in stock trading model is to take the right actions for trading at the right time, primarily based on the accurate forecast of future stock trends. Since an effective trading with given information of stock prices needs an intelligent strategy for the decision making, we applied Genetic Network Programming (GNP) to creating a stock trading model. In this paper, we propose a new method called Real Time Updating Genetic Network Programming (RTU-GNP) for adapting to the change of stock prices. There are three important points in this paper: First, the RTU-GNP method makes a stock trading decision considering both the recommendable information of technical indices and the candlestick charts according to the real time stock prices. Second, we combine RTU-GNP with a Sarsa learning algorithm to create the programs efficiently. Also, sub-nodes are introduced in each judgment and processing node to determine appropriate actions (buying/selling) and to select appropriate stock price information depending on the situation. Third, a Real Time Updating system has been firstly introduced in our paper considering the change of the trend of stock prices. The experimental results on the Japanese stock market show that the trading model with the proposed RTU-GNP method outperforms other models without real time updating. We also compared the experimental results using the proposed method with Buy&amp
    Hold method to confirm its effectiveness, and it is clarified that the proposed trading model can obtain much higher profits than Buy&amp
    Hold method. © 2009 The Institute of Electrical Engineers of Japan.

    DOI CiNii

  • Study of multi-branch structure of Universal Learning Networks

    S. Mabu, K. Shimada, K. Hirasawa, J. Hu

    Applied Soft Computing   9 ( 1 ) 393 - 403  2009

    DOI

  • Enhancing the generalization ability of neural networks through controlling the hidden layers

    Weishui Wan, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa, Jinglu Hu

    APPLIED SOFT COMPUTING   9 ( 1 ) 404 - 414  2009.01

     View Summary

    In this paper we proposed two new variants of backpropagation algorithm. The common point of these two new algorithms is that the outputs of nodes in the hidden layers are controlled with the aim to solve the moving target problem and the distributed weights problem. One algorithm (AlgoRobust) is not so insensitive to the noises in the data, the second one (AlgoGS) is through using Gauss-Schmidt algorithm to determine in each epoch which weight should be updated, while the other weights are kept unchanged in this epoch. In this way a better generalization can be obtained. Some theoretical explanations are also provided. In addition, simulation comparisons are made between Gaussian regularizer, optimal brain damage (OBD) and the proposed algorithms. Simulation results confirm that the new proposed algorithms perform better than that of Gaussian regularizer, and the first algorithm AlgoRobust performs better than the second algorithm AlgoGS in the noisy data. On the other hand AlgoGS performs better than the AlgoRobust on the data without noise and the final structure obtained by two new algorithms is comparable to that obtained by using OBD. (C) 2008 Elsevier B.V. All rights reserved.

    DOI

  • Genetic Network Programming with Rules

    F. Ye, L. Yu, S. Mabu, K. Shimada, K. Hirasawa

    Journal of advanced Computational Intelligence and Intelligent Informatics   13 ( 1 ) 16 - 24  2009

  • A Study of Double-Deck Elevator Systems Using Genetic Network Programming with Reinforcement Learning

    Z. Zhou, L. Yu, S. Mabu, K.Shimada, K. Hirasawa, S. Markon

    Journal of Advanced Computational Intelligence and Intelligent Informatics   13 ( 1 ) 35 - 43  2009

  • Q value-based dynamic programming with Boltzmann distribution for global optimal traffic routing strategy

    Shanqing Yu, Shingo Mabu, Fengming Ye, Hongqiang Wang, Kaoru Shimada, Kotaro Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   13 ( 5 ) 581 - 591  2009

     View Summary

    In this paper, we propose a heuristic method - Boltzmann Optimal Route Method trying to find a good approximation to the global optimum route for Origin- Destination pairs through iterations until the total traveling time converges. The overall idea of our method is to update the traveling time of each route section iteratively according to its corresponding traffic volume, and continuously generate a new global route by Q value-based Dynamic Programming combined with Boltzmann distribution. Finally, we can get the global optimum route considering the traffic volumes of the road sections. The new proposed method is compared with the conventional shortest-path method- Greedy strategy both in the static traffic system where the volumes of all the given Origin-Destination pairs of road networks are constant and in the dynamic traffic system in which changing traffic volumes are constantly provided. The results demonstrate that the proposed method performs better than the conventional method in global perspective.

    DOI

  • Combination of Two Evolutionary Methods for Mining Association Rules in Large and Dense Databases

    E. Gonzales, K. Taboada, S. Mabu, K. Shimada, K. Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   13 ( 5 ) 561 - 572  2009

  • Genetic Network Programming with Rule Accumulation and Its Application to Tile-WorldProblem

    L. Wang, S. Mabu, F. Ye, S. Eto, X. Fan, K. Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   13 ( 5 ) 551 - 560  2009

  • Traffic Flow Prediction with Genetic Network Programming(GNP)

    H. Zhou, S. Mabu, W. Wei, K. Shimada, K. Hirasawa

    Journal of Advanced Computaional Intelligence and Intelligent Informatics   13 ( 6 ) 713 - 725  2009

  • Global Optimal Routing Algorithm for Traffic Systems with Multiple ODs

    Y. Wang, S. Mabu, S. Eto, K Hirasawa

    Journal Advanced Computational Intelligence and Intelligent Informatics   13 ( 6 ) 704 - 712  2009

  • Genetic network programming for fuzzy association rule-based classification

    Karla Taboada, Shingo Mabu, Eloy Gonzales, Kaoru Shimada, Kotaro Hirasawa

    2009 IEEE Congress on Evolutionary Computation, CEC 2009   Norway   2387 - 2394  2009

     View Summary

    This paper presents a novel classification approach that integrates fuzzy classification rules and Genetic Network Programming (GNP). A fuzzy discretization technique is applied to transform the dataset, particularly for dealing with quantitative attributes. GNP is an evolutionary optimization technique that uses directed graph structures as genes instead of strings and trees of Genetic Algorithms (GA) and Genetic Programming (GP), respectively. This feature contributes to creating quite compact programs and implicitly memorizing past action sequences. Therefore, in the proposed method, taking the GNP's structure into account 1) extraction of fuzzy classification rules is done without identifying frequent itemsets used in most Apriori-based data mining algorithms, 2) calculation of the support, confidence and ?2 value is made in order to quantify the significance of the rules to be integrated into the classifier, 3) fuzzy membership values are used for fuzzy classification rules extraction, 4) fuzzy rules are mined through generations and stored in a general pool. On the other hand, parameters of the membership functions are evolved by non-uniform mutation in order to perform a more global search in the space of candidate membership functions. The performance of our algorithm has been compared with other relevant algorithms and the experimental results have shownthe advantages and effectiveness of the proposed model. © 2009 IEEE.

    DOI

  • Mining Multi-Class Datasets using Genetic Relation Algorithm for Rule Reduction

    E. Gonzales, K. Taboada, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2009   Norway   3249 - 3255  2009

    DOI

  • Multi-car elevator group supervisory control system using genetic network programming

    Lu Yu, Shingo Mabu, Tiantian Zhang, Shinji Eto, Kotaro Hirasawa

    2009 IEEE Congress on Evolutionary Computation, CEC 2009   Norway   2188 - 2193  2009

     View Summary

    Elevator group control systems are the transportation systems for handling passengers in the buildings. With the increasing demand for high-rise buildings, Multi- Car Elevator System(MCES) where two cars operate separately and independently in an elevator shaft are attracting attention as the next novel elevator system. Genetic Network Programming( GNP), one of the evolutionary computations, can realize a rule based MCES due to its directed graph structure of the individual, which makes the system more flexible. This paper discusses MCES using GNP for the buildings with 30 floors. The performance of MCES are examined and compared with Double-Deck Elevator System(DDES). © 2009 IEEE.

    DOI

  • Genetic Network Programming with Reconstructed Individuals

    F. Ye, S. Mabu, L. Wang, S. Eto, K. Hirasawa

    CEC 2009   Norway   854 - 859  2009

    DOI

  • Generalized Time Related Sequential Association Rule Mining and Traffic Prediction

    H. Zhou, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2009   Norway   2654 - 2661  2009

    DOI

  • Constructing Portfolio Investment Strategy Based on Time Adaptive Genetic Network Programming

    Y. Chen, S. Mabu, E. Ohkawa, K. Hirasawa

    CEC 2009   Norway   2379 - 2386  2009

    DOI

  • Genetic Network Programming with Rule Accumulation Considering Judgment Order

    L. Wang, F. Ye, S. Mabu, K. Hirasawa

    CEC 2009   Norway   3176 - 3182  2009

    DOI

  • Generalized Association Rules Mining with Multi-Branches Full-Paths and Its Application to Traffic Volume Prediction

    Huiyu Zhou, Shingo Mabu, Manoj Kanta Mainali, Xianneng Li, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   147 - 152  2009

  • Traveling Time Prediction using Isolation Rules

    Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009     1523 - 1528  2009

  • Robust Genetic Network Programming Using SARSA Learning for Autonomous Robots

    Sung Gil Park, Shingo Mabu, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   523 - 527  2009

  • Rule Accumulation Method with Modified Fitness Function based on Genetic Network Programming

    Lutao Wang, Shingo Mabu, Fengming Ye, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   1000 - 1005  2009

  • Temperature Parameter Control of Q value-based Dynamic Programming with Boltzmann Distribution

    Shanqing Yu, Shingo Mabu, Manoj Kanta Mainali, Shinji Eto, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   1471 - 1476  2009

  • Stock Movement Prediction using Fuzzy Intertransaction Class Association Rule Mining based on Genetic Network Programming

    Yuchen Yang, Shingo Mabu, Etsushi Ohkawa, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   2561 - 2566  2009

  • Global Portfolio Diversification by Genetic Relation Algorithm

    Victor Parque, Shingo Mabu, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   2567 - 2572  2009

  • A Genetic Relation Algorithm with Guided Mutation for the Large-Scale Portfolio Optimization

    Yan Chen, Chuan Yue, Shingo Mabu, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   2579 - 2584  2009

  • A New Associative Classification Method by Integrating CMAR and An Evolutionary Three-layers Structure

    Guangfei Yang, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   2920 - 2925  2009

  • Genetic Network Programming with Estimation of Distribution Algorithms, and its Application to Association Rule Mining for Traffic Prediction

    Xianneng Li, Shingo Mabu, Huiyu Zhou, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009     3457 - 3462  2009

  • Intrusion Detection System Combining Misuse Detection and Anomaly Detection Using Genetic Network Programming

    Yunlu Gong, Shingo Mabu, Ci Chen, Yifei Wang, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3463 - 3467  2009

  • Time Related Association Rules Mining for Traffic Prediction based on Genetic Network Programming combined with Estimation of Distribution Algorithms

    Yang Wang, Shingo Mabu, Huiyu Zhou, Xianneng Li, Kaoru Shimada

    ICROS-SICE International Joint Conference 2009   Fukuoka   3468 - 3473  2009

  • Genetic Network Programming with General Individual Reconstruction

    Fengming Ye, Shingo Mabu, Lutao Wang, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3474 - 3479  2009

  • Analysis of Fuzzy Class Association Rule Mining Based on Genetic Network Programming

    Ci Chen, Shingo Mabu, Chuan Yue, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3480 - 3484  2009

  • Global Optimal Routing for Traffic Systems with Multiple ODs using Genetic Algorithm

    Yu Wang, Shingo Mabu, Chuan Yue, Manoj Kanta Mainali, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3731 - 3737  2009

  • Q value-based Dynamic Programming with Evolving Penalties for Road Networks

    Manoj Kanta Mainali, Shingo Mabu, Yu Wang, Shanqing Yu, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3738 - 3743  2009

  • Class Association Rule Mining with Correlation Measures using Genetic Network Programming

    Eloy Gonzales, Shingo Mabu, Karla Taboada, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3850 - 3856  2009

  • Agent Bidding Strategy of Multiple Round English Auction based on Genetic Network Programming

    Chuan Yue, Shingo Mabu, Yan Chen, Yu Wang, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3857 - 3862  2009

  • Fuzzy Association Rule Mining and Classifier with Chi-squared Correlation Measure using Genetic Network Programming

    Karla Taboada, Shingo Mabu, Eloy Gonzales, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3863 - 3869  2009

  • Adaptive Controller for Double-Deck Elevator System using Genetic Network Programming

    Johanna Mansilla, Shingo Mabu, Lu Yu, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3870 - 3873  2009

  • A New Associative Classification Method by Integrating CMAR and RuleRank Model based on Genetic Network Programming

    Guangfei Yang, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   3874 - 3879  2009

  • Network intrusion detection using fuzzy class association rule mining based on genetic network programming

    Ci Chen, Shingo Mabu, Chuan Yue, Kaoru Shimada, Kotaro Hirasawa

    Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics   San Antonio Texas   60 - 67  2009

     View Summary

    Computer Systems are exposed to an increasing number and type of security threats due to the expanding of internet in recent years. How to detect network intrusions effectively becomes an important techniques. This paper presents a novel fuzzy class association rule mining method based on Genetic Network Programming(GNP) for detecting network intrusions. GNP is an evolutionary optimization techniques, which uses directed graph structures as genes instead of strings(Genetic Algorithm) or trees(Genetic Programming), leading to creating compact programs and implicitly memorizing past action sequences. By combining fuzzy set theory with GNP, the proposed method can deal with the mixed database which contains both discrete and continuous attributes. And it can be flexibly applied to both misuse and anomaly detection in Network Intrusion Detection Problem. Experimental results with KDD99Cup and DAPRA98 databases from MIT Lincoln Laboratory show that the proposed method provides a competitively high detection rate compared with other machine learning techniques. ©2009 IEEE.

    DOI

  • A study on energy consumption of elevator group supervisory control systems using genetic network programming

    Lu Yu, Shingo Mabu, Tiantian Zhang, Kotaro Hirasawa, Tsuyoshi Ueno

    Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics   San Antonio Texas   583 - 588  2009

     View Summary

    Elevator group supervisory control system (EGSCS) is a traffic system, where its controller manages the elevator movement to transport passengers in buildings efficiently. Recently, Artificial Intelligence (AI) technology has been used in such complex systems. Genetic Network Programming(GNP), a graph-based evolutionary method extended from GA and GP, has been already applied to EGSCS. On the other hand, since energy consumption is becoming one of the greatest challenges in the society, it should be taken as criteria of the elevator operations. Moreover, the elevator with maximum energy efficiency is therefore required. Finally, the simulations show that the elevator system has the higher energy consumption in the light traffic, thus, some factors have been introduced into GNP for energy saving in this paper. ©2009 IEEE.

    DOI

  • Backward time related association rule mining with database rearrangement in traffic volume prediction

    Huiyu Zhou, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics   San Antonio Texas   1021 - 1026  2009

     View Summary

    In this paper, Backward Time Related Association Rule Mining using Genetic Network Programming (GNP) with Database Rearrangement is introduced in order to find time related sequential association from time related databases effectively and efficiently. GNP is a kind of human brain like evolutionary model which represents solutions as directed graph structures. The concept of database rearrangement to better handle association rule extraction from the databases in the traffic volume prediction problems is proposed. The proposed algorithm and experimental results are also included. ©2009 IEEE.

    DOI

  • Evolving Plural Programs by Genetic Network Programming with Multi-Start Nodes

    S. Mabu, K. Hirasawa

    IEEE SMC 2009   San Antonio Texas  2009

    DOI

  • Fuzzy classification rule mining based on genetic network programming algorithm

    Karla Taboada, Shingo Mabu, Eloy Gonzales, Kaoru Shimada, Kotaro Hirasawa

    Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics   San Antonio Texas   3860 - 3865  2009

     View Summary

    Association rule-based classification is one of the most important data mining techniques applied to many scientific problems. In the last few years, extensive research has been carried out to develop enhanced methods and obtained higher classification accuracies than traditional classifiers. However, the current studies show that the association rule-based classifiers may also suffer some problems inherited from association rule mining such as handling of (1) continuous data and (2) the support/confidence framework. In this paper, a novel fuzzy classification model based on Genetic Network Programming (GNP) that can deal with the above problems has been proposed. GNP is one of the evolutionary optimization algorithms that uses directed graph structures as solutions instead of strings (Genetic Algorithms) or trees (Genetic Programming). Therefore, GNP can deal with more complex problems by using the higher expression ability of graph structures. The performance of our algorithm has been compared with other relevant algorithms and the experimental results show the advantages and effectiveness of the proposed model. ©2009 IEEE.

    DOI

  • Multi-Route Algorithm using Temperature Control of Boltzmann Distribution in Q Value-based Dynamic Programming

    S. Yu, S. mabu, M. K. Mainali, S. Eto, K. Shimada, K. Hirasawa

    IEEE SMC 2009   San Antonio Texas  2009

    DOI

  • A portfolio selection model using genetic relation algorithm and genetic network programming

    Yan Chen, Shingo Mabu, Kotaro Hirasawa

    Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics   San Antonio Texas   4378 - 4383  2009

     View Summary

    In this paper, a new evolutionary method named genetic relation algorithm (GRA) has been proposed and applied to the portfolio selection problem. The number of brands in the stock market is generally very large, therefore, techniques for selecting the effective portfolio are likely to be of interest in the financial field. In order to pick up a fixed number of the most efficient portfolio, the proposed model considers the correlation coefficient between stocks as strength, which indicates the relationship between nodes in GRA. The algorithm evaluates the relationships between stock brands using a specific measure of strength and generates the optimal portfolio in the final generation. The efficiency of GRA method is confirmed by the stock trading model using genetic network programming (GNP) that has been proposed in the previous study. We present the experimental results obtained by GRA and compare them with those obtained by traditional method, and it is clarified that the proposed model can obtain much higher profits than the traditional one. ©2009 IEEE.

    DOI

  • Real time Updating Genetic Network Programming for adapting to the change of stock prices

    Yan Chen, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   129 ( 2 ) 22 - 354  2009

     View Summary

    The key in stock trading model is to take the right actions for trading at the right time, primarily based on the accurate forecast of future stock trends. Since an effective trading with given information of stock prices needs an intelligent strategy for the decision making, we applied Genetic Network Programming (GNP) to creating a stock trading model. In this paper, we propose a new method called Real Time Updating Genetic Network Programming (RTU-GNP) for adapting to the change of stock prices. There are three important points in this paper: First, the RTU-GNP method makes a stock trading decision considering both the recommendable information of technical indices and the candlestick charts according to the real time stock prices. Second, we combine RTU-GNP with a Sarsa learning algorithm to create the programs efficiently. Also, sub-nodes are introduced in each judgment and processing node to determine appropriate actions (buying/selling) and to select appropriate stock price information depending on the situation. Third, a Real Time Updating system has been firstly introduced in our paper considering the change of the trend of stock prices. The experimental results on the Japanese stock market show that the trading model with the proposed RTU-GNP method outperforms other models without real time updating. We also compared the experimental results using the proposed method with Buy&amp
    Hold method to confirm its effectiveness, and it is clarified that the proposed trading model can obtain much higher profits than Buy&amp
    Hold method. © 2009 The Institute of Electrical Engineers of Japan.

    DOI CiNii

  • Study of multi-branch structure of Universal Learning Networks

    Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa, Jinglu Hu

    APPLIED SOFT COMPUTING   9 ( 1 ) 393 - 403  2009.01

     View Summary

    In this paper, multi-branch structure of Universal Learning Networks (ULNs) is studied to verify its effectiveness for obtaining compact models, which have neurons connected with other neurons using more than two branches having nonlinear functions. Multi-branch structure has been proved to have higher representation/generalization ability and lower computational cost than conventional neural networks because of the nonlinear function of the multi-branches and the reduction of the number of neurons to be used. In addition, learning of delay elements of multi-branch ULNs has improved their potential to build up a compact dynamical model with higher performances and lower computational cost when applied for identifying dynamical systems. (C) 2008 Elsevier B. V. All rights reserved.

    DOI

  • Enhancing the generalization ability of neural networks through controlling the hidden layers

    Weishui Wan, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa, Jinglu Hu

    APPLIED SOFT COMPUTING   9 ( 1 ) 404 - 414  2009.01

     View Summary

    In this paper we proposed two new variants of backpropagation algorithm. The common point of these two new algorithms is that the outputs of nodes in the hidden layers are controlled with the aim to solve the moving target problem and the distributed weights problem. One algorithm (AlgoRobust) is not so insensitive to the noises in the data, the second one (AlgoGS) is through using Gauss-Schmidt algorithm to determine in each epoch which weight should be updated, while the other weights are kept unchanged in this epoch. In this way a better generalization can be obtained. Some theoretical explanations are also provided. In addition, simulation comparisons are made between Gaussian regularizer, optimal brain damage (OBD) and the proposed algorithms. Simulation results confirm that the new proposed algorithms perform better than that of Gaussian regularizer, and the first algorithm AlgoRobust performs better than the second algorithm AlgoGS in the noisy data. On the other hand AlgoGS performs better than the AlgoRobust on the data without noise and the final structure obtained by two new algorithms is comparable to that obtained by using OBD. (C) 2008 Elsevier B.V. All rights reserved.

    DOI

  • Genetic Network Programming with Rules

    F. Ye, L. Yu, S. Mabu, K. Shimada, K. Hirasawa

    Journal of advanced Computational Intelligence and Intelligent Informatics   13 ( 1 ) 16 - 24  2009

  • A Study of Double-Deck Elevator Systems Using Genetic Network Programming with Reinforcement Learning

    Z. Zhou, L. Yu, S. Mabu, K.Shimada, K. Hirasawa, S. Markon

    Journal of Advanced Computational Intelligence and Intelligent Informatics   13 ( 1 ) 35 - 43  2009

  • A nonlinear model to rank association rules based on semantic similarity and genetic network programing

    Guangfei Yang, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa

    IEEJ Transactions on Electrical and Electronic Engineering   4 ( 2 ) 248 - 256  2009

     View Summary

    Many methods have been studied for mining association rules efficiently. However, because these methods usually generate a large number of rules, it is still a heavy burden for the users to find the most interesting ones. In this paper, we propose a novel method for finding what the user is interested in by assigning several keywords, like searching documents on the Web using search engines. By considering both the semantic similarity between the rules and keywords, and the statistical information like support, confidence, chi-squared value, etc. we could rank the rules by a new method named RuleRank, where evolutionary methods are applied to find the optimal ranking model. Experiments show that our approach is effective for the users to find what they want. © 2009 Institute of Electrical Engineers of Japan.

    DOI CiNii

  • Q value-based dynamic programming with Boltzmann distribution for global optimal traffic routing strategy

    Shanqing Yu, Shingo Mabu, Fengming Ye, Hongqiang Wang, Kaoru Shimada, Kotaro Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   13 ( 5 ) 581 - 591  2009

     View Summary

    In this paper, we propose a heuristic method - Boltzmann Optimal Route Method trying to find a good approximation to the global optimum route for Origin- Destination pairs through iterations until the total traveling time converges. The overall idea of our method is to update the traveling time of each route section iteratively according to its corresponding traffic volume, and continuously generate a new global route by Q value-based Dynamic Programming combined with Boltzmann distribution. Finally, we can get the global optimum route considering the traffic volumes of the road sections. The new proposed method is compared with the conventional shortest-path method- Greedy strategy both in the static traffic system where the volumes of all the given Origin-Destination pairs of road networks are constant and in the dynamic traffic system in which changing traffic volumes are constantly provided. The results demonstrate that the proposed method performs better than the conventional method in global perspective.

    DOI

  • Combination of Two Evolutionary Methods for Mining Association Rules in Large and Dense Databases

    E. Gonzales, K. Taboada, S. Mabu, K. Shimada, K. Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   13 ( 5 ) 561 - 572  2009

  • Genetic Network Programming with Rule Accumulation and Its Application to Tile-WorldProblem

    L. Wang, S. Mabu, F. Ye, S. Eto, X. Fan, K. Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   13 ( 5 ) 551 - 560  2009

  • Traffic Flow Prediction with Genetic Network Programming(GNP)

    H. Zhou, S. Mabu, W. Wei, K. Shimada, K. Hirasawa

    Journal of Advanced Computaional Intelligence and Intelligent Informatics   13 ( 6 ) 713 - 725  2009

  • Global Optimal Routing Algorithm for Traffic Systems with Multiple ODs

    Y. Wang, S. Mabu, S. Eto, K Hirasawa

    Journal Advanced Computational Intelligence and Intelligent Informatics   13 ( 6 ) 704 - 712  2009

  • A Traffic-Flow-Adaptive Controller of Double-Deck Elevator Systems using Genetic Network Programming

    Jin Zhou, Lu Yu, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa, Sandor Markon

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   3 ( 6 ) 703 - 714  2008.11

     View Summary

    The double-deck elevator system (DDES) has been invented firstly as a solution to improve the transportation capacity of elevator group systems in the up-peak traffic pattern. The transportation capacity could be even doubled when DDES runs in a pure up-peak traffic pattern where two connected cages stop at every two floors in an elevator round trip. However, the specific features of DDES make the elevator system intractable when it runs in some other traffic patterns. Moreover, since almost all the traffic flows vary continuously during a day, an optimized controller of DDES is required to adapt to the varying traffic flow. In this paper, we have proposed a controller adaptive to traffic flows for DDES using Genetic Network Programming (GNP) based on Our past studies in this field, where the effectiveness of DDES controller using GNP has been verified in three typical traffic patterns. A new traffic flow judgment part was introduced into the GNP framework of DDES controller in this paper, and the different parts of GNP were expected to be functionally localized by the evolutionary process to make the appropriate cage assignment in different traffic flow patterns. Simulation results show that the proposed method outperforms a conventional approach and two heuristic approaches in a varying traffic flow during the working time of a typical office building. (C) 2008 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI

  • A Traffic-Flow-Adaptive Controller of Double-Deck Elevator Systems using Genetic Network Programming

    Jin Zhou, Lu Yu, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa, Sandor Markon

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   3 ( 6 ) 703 - 714  2008.11

     View Summary

    The double-deck elevator system (DDES) has been invented firstly as a solution to improve the transportation capacity of elevator group systems in the up-peak traffic pattern. The transportation capacity could be even doubled when DDES runs in a pure up-peak traffic pattern where two connected cages stop at every two floors in an elevator round trip. However, the specific features of DDES make the elevator system intractable when it runs in some other traffic patterns. Moreover, since almost all the traffic flows vary continuously during a day, an optimized controller of DDES is required to adapt to the varying traffic flow. In this paper, we have proposed a controller adaptive to traffic flows for DDES using Genetic Network Programming (GNP) based on Our past studies in this field, where the effectiveness of DDES controller using GNP has been verified in three typical traffic patterns. A new traffic flow judgment part was introduced into the GNP framework of DDES controller in this paper, and the different parts of GNP were expected to be functionally localized by the evolutionary process to make the appropriate cage assignment in different traffic flow patterns. Simulation results show that the proposed method outperforms a conventional approach and two heuristic approaches in a varying traffic flow during the working time of a typical office building. (C) 2008 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI

  • A double-deck elevator group supervisory control system using genetic network programming

    Kotaro Hirasawa, Toru Eguchi, Jin Zhou, Lu Yu, Jinglu Hu, Sandor Markon

    IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART C-APPLICATIONS AND REVIEWS   38 ( 4 ) 535 - 550  2008.07

     View Summary

    Elevator group supervisory control systems (EGSCSs) are designed so that the movement of several elevators in a building is controlled efficiently. The efficient control of EGSCSs using conventional control methods is very difficult due to its complexity, so it is becoming popular to introduce artificial intelligence (AI) technologies into EGSCSs in recent years. As a new approach, a graph-based evolutionary method named genetic network programming (GNP) has been applied to the EGSCSs, and its effectiveness is clarified. The GNP can introduce various a priori knowledge of the EGSCSs in its node functions easily, and can execute an efficient rule-based group supervisory control that is optimized in an evolutionary way. Meanwhile, double-deck elevator systems (DDESs) where two cages are connected in a shaft have been developed for the rising demand of more efficient transport of passengers in high-rise buildings. The DDESs have specific features due to the connection of cages and the need for comfortable riding; so its group supervisory control becomes more complex and requires more efficient group control systems than the conventional single-deck elevator systems (SDESs). In this paper, a new group supervisory control system for DDESs using GNP is proposed, and its optimization and performance evaluation are done through simulations. First, optimization of the GNP for DDSEs is executed. Second, the performance of the proposed method is evaluated by comparison with conventional methods, and the obtained control rules in GNP are studied. Finally, the reduction of space requirements compared with SDESs is confirmed.

    DOI CiNii

  • A double-deck elevator group supervisory control system using genetic network programming

    Kotaro Hirasawa, Toru Eguchi, Jin Zhou, Lu Yu, Jinglu Hu, Sandor Markon

    IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART C-APPLICATIONS AND REVIEWS   38 ( 4 ) 535 - 550  2008.07

     View Summary

    Elevator group supervisory control systems (EGSCSs) are designed so that the movement of several elevators in a building is controlled efficiently. The efficient control of EGSCSs using conventional control methods is very difficult due to its complexity, so it is becoming popular to introduce artificial intelligence (AI) technologies into EGSCSs in recent years. As a new approach, a graph-based evolutionary method named genetic network programming (GNP) has been applied to the EGSCSs, and its effectiveness is clarified. The GNP can introduce various a priori knowledge of the EGSCSs in its node functions easily, and can execute an efficient rule-based group supervisory control that is optimized in an evolutionary way. Meanwhile, double-deck elevator systems (DDESs) where two cages are connected in a shaft have been developed for the rising demand of more efficient transport of passengers in high-rise buildings. The DDESs have specific features due to the connection of cages and the need for comfortable riding; so its group supervisory control becomes more complex and requires more efficient group control systems than the conventional single-deck elevator systems (SDESs). In this paper, a new group supervisory control system for DDESs using GNP is proposed, and its optimization and performance evaluation are done through simulations. First, optimization of the GNP for DDSEs is executed. Second, the performance of the proposed method is evaluated by comparison with conventional methods, and the obtained control rules in GNP are studied. Finally, the reduction of space requirements compared with SDESs is confirmed.

    DOI CiNii

  • Association rule mining for continuous attributes using genetic network programming

    Karla Taboada, Eloy Gonzales, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   3 ( 2 ) 199 - 211  2008.03

     View Summary

    Most of the existing association rule mining algorithms are able to extract knowledge from databases with attributes of binary values. However, in real-world applications, databases are usually composed of continuous values such as height, length or weight. If the attributes are continuous, the algorithms are commonly integrated with a discretization method that transforms them into discrete attributes. Discretization is a process of transforming a continuous attribute value into a finite number of intervals and assigning each interval into a discrete numerical value. However, the user most often must specify the number of intervals, or provide some heuristic rules to be used while discretization, and then it is difficult to get the highest attribute interdependency and at the same time get the lowest number of intervals. In this paper we present an association rule mining algorithm that is suited for continuous valued attributes commonly found in scientific and statistical databases. We propose a method using a new graph-based evolutionary algorithm named 'genetic network programming (GNP)' that can deal with continuous values directly, that is, without using any discretization method as a preprocessing step. GNP represents its individuals using graph structures and evolves them in order to find a solution; this feature contributes to creating very compact programs and implicitly memorizing past action sequences. In the proposed method using GNP, the significance of the extracted association rules is measured by the use of chi(2) test, and only important association rules are stored in a pool all together through generations. Results of experiments conducted on a real-life database suggest that the proposed method provides an effective technique for handling continuous attributes. (C) 2008 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • Association rule mining for continuous attributes using genetic network programming

    Karla Taboada, Eloy Gonzales, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu

    IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING   3 ( 2 ) 199 - 211  2008.03

     View Summary

    Most of the existing association rule mining algorithms are able to extract knowledge from databases with attributes of binary values. However, in real-world applications, databases are usually composed of continuous values such as height, length or weight. If the attributes are continuous, the algorithms are commonly integrated with a discretization method that transforms them into discrete attributes. Discretization is a process of transforming a continuous attribute value into a finite number of intervals and assigning each interval into a discrete numerical value. However, the user most often must specify the number of intervals, or provide some heuristic rules to be used while discretization, and then it is difficult to get the highest attribute interdependency and at the same time get the lowest number of intervals. In this paper we present an association rule mining algorithm that is suited for continuous valued attributes commonly found in scientific and statistical databases. We propose a method using a new graph-based evolutionary algorithm named 'genetic network programming (GNP)' that can deal with continuous values directly, that is, without using any discretization method as a preprocessing step. GNP represents its individuals using graph structures and evolves them in order to find a solution; this feature contributes to creating very compact programs and implicitly memorizing past action sequences. In the proposed method using GNP, the significance of the extracted association rules is measured by the use of chi(2) test, and only important association rules are stored in a pool all together through generations. Results of experiments conducted on a real-life database suggest that the proposed method provides an effective technique for handling continuous attributes. (C) 2008 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

    DOI CiNii

  • genetic Network Programming with Rule Accumulation

    L. Wang, C. Chen, S. Mabu, K. Shimada, K. Hirasawa

    SICE SSI 2008    2008

  • Global Optimal routing Algorithm for Traffic System with Multiple ODs

    Y. Wang, S. Mabu, K. Shimada, K. Hirasawa

    FAN 2008    2008

  • Multiple Round English Auction based on Genetic Network Programming

    C. Yue, Y. Wang, S. Mabu, K. Shimada, K. Hirasawa

    FAN 2008    2008

  • Class Association Rule Mining using Genetic Network Programming for Stock Market Prediction

    Y. Yang, Y. Chen, E. Ohkawa, K. Shimaad, S. Mabu, K. Hirasawa

    FAN 2008    2008

  • Anomaly Detection Using Class Association Rule Mining Based on Genetic Network Programming

    C. Chen, L. Wang, S. Mabu, K. Shimada, K. Hirasawa

    FAN 2008    2008

  • Adaptive Training System for Double Deck Elevator System using Genetic Network Programming

    J. Mansilla, J. Zhou, L. Yu, S. Mabu, K. Shimada, K. Hirasawa

    FAN 2008    2008

  • Real Time Updating Genetic Network Programming for Adapting the Change of Stock Prices

    Y. Chen, S. Mabu, K. Shimada, K. Hirasawa

    CEC 2008   Hong Kong   370 - 377  2008

    DOI

  • Comparative Association Rules Mining using Genetic Network Programming(GNP) with attribute Accumulation Mechanism and Its Application to Traffic Systems

    W. Wei, H. Zhou, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2008   Hong Kong   292 - 298  2008

    DOI

  • Double-Deck Elevator Systems Adaptive to Traffic Flows using Genetic Network Programming

    J. Zhou, L. Yu, S. Mabu, K. Shimada, K. Hirasawa, S. Markon

    CEC 2008   Hong Kong   773 - 778  2008

    DOI

  • Evaluating Class Association Rules using Genetic Relation Programming

    E. Gonzales, K. Taboada, K. Shimada, S. mabu, K. Hirasawa

    CEC 2008   Hong Kong   731 - 736  2008

    DOI

  • A Personalaized Association Rule Ranking Method Based on Semantic Similarity and Evolutionary Computation

    G. Yang, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2008   Hong Kong   487 - 494  2008

    DOI

  • Genetic network programming with rules

    Fengming Ye, Shigo Mabu, Kaoru Shimada, Kotaro Hirasawa

    2008 IEEE Congress on Evolutionary Computation, CEC 2008   Hong Kong   413 - 418  2008

     View Summary

    Genetic Network Programming (GNP) is an evolutionary approach which can evolve itself and find the optimal solutions. As many papers have demonstrated that GNP which has a directed graph structure can deal with dynamic environments very efficiently and effectively. It can be used in many areas such as data mining, forecasting stock markets, elevator system problems, etc. In order to improve GNP's performance further, this paper proposes a method called GNP with Rules. The aim of the proposal method is to balance exploitation and exploration, that is, to strengthen exploitation ability by using the exploited information extensively during the evolution process of GNP. The proposal method consists of 4 steps: rule extraction, rule selection, individual reconstruction and individual replacement. Tile-world was used as a simulation environment. The simulation results show some advantages of GNP with Rules over conventional GNPs. © 2008 IEEE.

    DOI

  • Time Related Association Rule Mining with Attribute Accumulation Mechanism and Its Application to Traffic Prediction

    H. Zhou, W. Wei, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2008   Hong Kong   305 - 311  2008

    DOI

  • Genetic Network Programming Based Data Mining Method for Extracting Fuzzy Association Rules

    K. Taboada, E. Gonzales, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2008     1756 - 1763  2008

    DOI

  • Optimal Route of Road Networks by Dynamic Programming

    M.K. Mainali, K. Shimada, S. Mabu, K. Hirasawa

    IJCNN 2008     3416 - 3420  2008

    DOI

  • Construction of Portfolio Optimization System using Genetic Network Programming with Control Nodes

    Y. Chen, S. Mabu, K. Shimada, K. Hirasawa

    GECCO 2008   Atlanta   1693 - 1694  2008

  • Stock Trading Strategies by Genetic Network Programming with Flag Nodes

    S. Mabu, Y. Chen, E. Ohkawa, K. Hirasawa

    GECCO 2008   Atlanta   1709 - 1710  2008

  • Varying Portfolio Construction of Stocks Using Genetic Network Programming with Control Nodes

    E. Ohkawa, Y. Chen, S. Mabu, K. Shimada, K. Hirasawa

    GECCO 2008   Atlanta   1715 - 1716  2008

  • double-Deck Elevator System Using Genetic Network Programming with Genetic Operators based on Pheromone Information

    L. Yu, J. Zhou, F. Ye, S. Mabu, K. Shimada, K. Hirasawa

    GECCO 2008   Atlanta   2239 - 2244  2008

  • Exceptional Association Rule Mining Using Genetic Network Programming

    SHIMADA K.

    Proc. of the 4th International Conference on Data Mining (DMIN 2008)   Las Vegas   277 - 283  2008

    CiNii

  • A Nonlinear Semantic Model for Selecting Association Rules for Users

    G. Yang, K. Shimada, S. Mabu, K. Hirasawa

    SICE 2008   Tokyo   121 - 126  2008

  • Multi-Car Elevator System using Genetic Network Programming

    L. Yu, J. zhou, S. Mabu, K. Shimada, K. Hirasawa, S. Markon

    SICE 2008   Tokyo   127 - 131  2008

    DOI

  • Multi-objective Optimal Rout Search for Road Networks by Dynamic Programming

    M. Mainali, K. Shimada, S. Mabu, K. Hirasawa

    SICE 2008   Tokyo   628 - 632  2008

  • A stock trading model for multi-brands optimization based on genetic network programming with control nodes

    Yan Chen, Etsushi Ohkawa, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    Proceedings of the SICE Annual Conference   Tokyo   664 - 669  2008

     View Summary

    Many evolutionary computation methods applied to the financial field have been reported. A new evolutionary method named "Genetic Network Programming" (GNP) has been developed and applied to the stock market recently. The efficient trading rules created by GNP has been confirmed in our previous research. In this paper a Multi-Brands optimization system based on Genetic Network Programming with control nodes is presented. This method makes use of the information from Technical Indices and Candlestick Chart. The proposed optimization system, consisting of technical analysis rules, are trained to generate trading advice. The experimental results on the Japanese stock market show that the proposed optimization system using GNP with control nodes method outperforms other traditional models in terms of both accuracy and efficiency. © 2008 SICE.

    DOI

  • Traffic flow prediction with genetic network programming

    Wei Wei, Huiyu Zhou, Manoj Kanta Mainali, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa

    Proceedings of the SICE Annual Conference   Tokyo   670 - 675  2008

     View Summary

    In this paper, a method for traffic flow prediction has been proposed to obtain prediction rules from the past traffic data using Genetic Network Programming(GNP). GNP is an evolutionary approach which can evolve itself and find the optimal solutions. It has been clarified that GNP works well especially in dynamic environments since GNP is consisted of graphic structures, creates quite compact programs and has an implicit memory function. In this paper, GNP is applied to create a traffic flow prediction model. And we proposed the spatial adjacency model for the prediction and two kinds of models for N-step prediction. Additionally, the adaptive penalty functions are adopted for the fitness function in order to alleviate the infeasible solutions containing loops in the training process. Furthermore, the sharing function is also used to avoid the premature convergence. © 2008 SICE.

    DOI

  • Class Association Rules Mining with Time Series and Its Application to Traffic Load prediction

    H. Zhou, W. Wei, M. Mainali, K. Shimada, S. Mabu, K. Hirasawa

    SICE 2008   Tokyo   1187 - 1192  2008

  • Solving Multi-objective Optimization Problem by RasID-GA

    M. Ogata, D. Sohn, S. Mabu, K. Shimada, K. Hirasawa

    SICE 2008   Tokyo   1193 - 1198  2008

  • Discovering fuzzy classification rules using genetic network programming

    Karla Taboada, Eloy Gonzales, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa

    Proceedings of the SICE Annual Conference   Tokyo   1788 - 1793  2008

     View Summary

    Classification rule mining is an active data mining research area. Most related studies have shown how binary valued datasets are handled. However, datasets in real-world applications, usually consist of fuzzy and quantitative values. As a result, the idea to combine the different approaches with Fuzzy Set Theory has been applied more frequently in recent years. Fuzzy sets can help to overcome the so-called sharp boundary problem by allowing partial memberships to the different sets, not only 1 and 0. On the other hand, Fuzzy Sets Theory has been shown to be a very useful tool because the mined rules are expressed in linguistic terms, which are more natural and understandable for human beings. This paper proposes the combination of Fuzzy set theory and "Genetic Network Programming" (GNP) for discovering fuzzy classification rules from given quantitative data. GNP, as an extension of Genetic Algorithms (GA) and Genetic Programming (GP), is an evolutionary optimization technique that uses directed graph structures as genes instead of strings and trees
    this feature contributes creating quite compact programs and implicitly memorizing past action sequences. At last, experimental results conducted on a real world database verify the performance of the proposed method. © 2008 SICE.

    DOI

  • A Q Value-based Dynamic Programming Algorithm with Boltzmann Distribution for Optimizing the Global Traffic Routing Strategy

    S. Yu, H. Wang, F. Ye, S. Mabu, K. Shimada, K. Hirasawa

    SICE 2008   Tokyo   619 - 622  2008

  • A Global Routing Strategy in Dynamic Traffic Environments with a Combination of Q Value-based Dynamic Programming and Boltzmann Distribution

    S. Yu, F. Ye, H. Wang, S. Mabu, K. Shimada, S. Yu, K. Hirasawa

    SICE 2008   Tokyo   623 - 627  2008

    DOI

  • Evaluation of varying portfolio construction of stocks using genetic network programming with control nodes

    Etsushi Ohkawa, Yan Chen, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    Proceedings of the SICE Annual Conference   Tokyo   1231 - 1236  2008

     View Summary

    A new evolutionary method named "Genetic Network Programming with control nodes, GNPcn" has been applied to determine the timing of buying or selling stocks. GNPcn represents its solutions as directed graph structures which has some useful features inherently. For example, GNPcn has an implicit memory function which memorizes the past action sequences of agents and GNPcn can re-use nodes repeatedly in the network flow, so very compact graph structures can be made. GNPcn can determine the strategy of buying and selling stocks of multi issues. And GNPcn can distribute the purchase capital to each stock based on the distribution ratio. The effectiveness of the proposed method is confirmed by simulations. © 2008 SICE.

    DOI

  • Idle Cage Assignment Algorithm-embedded Controller of Doube-deck Elevator Systems using Genetic Network Programming

    J, Zhou, L. Yu, S. Mabu, K. Shimada, K. Hirasawa, S. Markon

    SICE 2008   Tokyo   146 - 150  2008

    DOI

  • Genetic network programming with rule chains

    Fengming Ye, Shigo Mabu, Kaoru Shimada, Kotaro Hirasawa

    Proceedings of the SICE Annual Conference   Tokyo   1220 - 1225  2008

     View Summary

    Genetic Network Programming (GNP) is a newly developed evolutionary approach which can evolve itself and And the optimal solutions. A lot of research has been done and it has been demonstrated that GNP which has a directed graph structure can deal with dynamic environments very efficiently and effectively. It can be used in many areas such data mining, elevator supervising control systems, the strategy of buying and selling stocks in stock markets, forecasting the traffic volumes in road networks, etc. In order to improve GNP's performance further, this paper proposes a method called GNP with Rule Chains. The aim of the proposed method is to balance exploitation and exploration, that is, to strengthen exploitation ability by using the exploited information extensively during the evolution process of GNP. The proposed method consists of 4 steps: rule extraction, rule selection, individual reconstruction and individual replacement. Tileworld was used as a simulation environment. The simulation results show some advantages of GNP with Rule Chains over conventional GNPs. © 2008 SICE.

    DOI

  • Performance Evaluation of Genetic Network Programming with Actor-Critic for Creating Mobile Robot Behavior

    MABU Shingo, HIRASAWA Kotaro, HATAKEYAMA Hiroyuki, FURUZUKI Takayuki

    Transactions of the Society of Instrument and Control Engineers   44 ( 4 ) 343 - 350  2008

     View Summary

    Genetic Network Programming (GNP) has been proposed as a new graph-based evolutionary algorithm. GNP represents its solutions as graph structures which contribute to improving the expression ability of the programs. GNP with Reinforcement Learning (GNP-RL) was also proposed as an extended algorithm of GNP and its effectiveness has been confirmed. Because GNP-RL executes reinforcement learning during task execution in addition to evolution after task execution, it can search for solutions efficiently. In this paper, GNP with Actor-Critic (GNP-AC) is proposed to enhance the effectiveness of GNP-RL. Actor-Critic can adjust numerical values appropriately during task execution, i. e., online learning, and use them for determining actions. To confirm the effectiveness of the proposed method, GNP-AC is applied to the controller of the Khepera simulator and its generalization ability is evaluated.

    DOI CiNii

  • 遺伝的ネットワークプログラミングによる不完全データベースからのクラス相関ルールの抽出

    嶋田香, 間普真吾, 森川英治, 平澤宏太郎, 古月敬之

    電気学会論文集 C   128 ( 5 ) 795 - 803  2008

    DOI CiNii

  • Genetic Network Programming with Intron-Like Nodes

    MABU Shingo, CHEN Yan, ETO Shinji, SHIMADA Kaoru, HIRASAWA Kotaro

    IEEJ Transactions on Electronics, Information and Systems   128 ( 8 ) 1312 - 1319  2008

     View Summary

    Recently, Genetic Network Programming (GNP) has been proposed, which is an extension of Genetic Algorithm(GA) and Genetic Programming(GP). GNP can make compact programs and can memorize the past history in it implicitly, because it expresses the solution by directed graphs and therefore, it can reuse the nodes. In this research, intron-like nodes are introduced for improving the performance of GNP. The aim of introducing intron-like nodes is to use every node as much as possible. It is found from simulations that the intron-like nodes are useful for improving the training speed and generalization ability.

    DOI CiNii

  • フラグノードおよび重要度指標の調整を用いた遺伝的ネットワークプログラミングによる株式売買ルールの生成

    間普真吾, 陳艶, 平澤幸太郎

    電気学会論文誌 C   128 ( 9 ) 1462 - 1469  2008

    DOI CiNii

  • Buying and selling stocks of multi brands using Genetic Network Programming with control nodes

    Etsushi Ohkawa, Yan Chen, Zhiguo Bao, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   128 ( 12 ) 13 - 1819  2008

     View Summary

    A new evolutionary method named "Genetic Network Programming with control nodes, GNPcn" has been applied to determine the timing of buying or selling stocks. GNPcn represents its solutions as directed graph structures which has some useful features inherently. For example, GNPcn has an implicit memory function which memorizes the past action sequences of agents and GNPcn can re-use nodes repeatedly in the network flow, so very compact graph structures can be made. GNPcn can determine the strategy of buying and selling stocks of multi issues. The effectiveness of the proposed method is confirmed by simulations. © 2008 The Institute of Electrical Engineers of Japan.

    DOI CiNii

  • A Global Optimization Method RasID-GA for Neural Network Training

    D. Sohn, S. Mabu, K. Shimada, K. Hirasawa, J. Hu

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 1 ) 85 - 93  2008

  • A Genetic Network Programming Based Method to Mine Generalized Association Rules with Ontology

    G. Yang, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 1 ) 63 - 76  2008

  • Support Vector Machine Classifier with WHM Offset for Unbalanced Data

    B. Li, J. Hu, K. Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 1 ) 94 - 101  2008

  • Trading rules on stock markets using genetic network programming with Sarsa learning

    CHEN Y

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 4 ) 383 - 392  2008

    CiNii

  • Comparative Association Rules Mining Using Genetic Network Programming(GNP) with Attributes Accumulation Mechanism and its Application to Traffic systems

    W. Wei, H. Zhou, K. Shimada, S. Mabu, K. Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 4 ) 393 - 403  2008

  • Time Related Association Rules Mining with Attribute Accumulation Mechanism and Its Application to Traffic Prediction

    H. Zhou, W. Wei, K. Shimada, S. Mabu, K. Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 5 ) 467 - 478  2008

  • Optimal Route Based on Dynamic Programming for Road networks

    M. K. Mainali, K. Shimada, S. Mabu, K. Hirasawa

    Journal of Advanced Computer Intelligence and Intelligent Informatics   12 ( 6 ) 546 - 553  2008

  • genetic Network Programming with Rule Accumulation

    L. Wang, C. Chen, S. Mabu, K. Shimada, K. Hirasawa

    SICE SSI 2008    2008

  • Global Optimal routing Algorithm for Traffic System with Multiple ODs

    Y. Wang, S. Mabu, K. Shimada, K. Hirasawa

    FAN 2008    2008

  • Multiple Round English Auction based on Genetic Network Programming

    C. Yue, Y. Wang, S. Mabu, K. Shimada, K. Hirasawa

    FAN 2008    2008

  • Class Association Rule Mining using Genetic Network Programming for Stock Market Prediction

    Y. Yang, Y. Chen, E. Ohkawa, K. Shimaad, S. Mabu, K. Hirasawa

    FAN 2008    2008

  • Anomaly Detection Using Class Association Rule Mining Based on Genetic Network Programming

    C. Chen, L. Wang, S. Mabu, K. Shimada, K. Hirasawa

    FAN 2008    2008

  • Adaptive Training System for Double Deck Elevator System using Genetic Network Programming

    J. Mansilla, J. Zhou, L. Yu, S. Mabu, K. Shimada, K. Hirasawa

    FAN 2008    2008

  • Real Time Updating Genetic Network Programming for Adapting the Change of Stock Prices

    Y. Chen, S. Mabu, K. Shimada, K. Hirasawa

    CEC 2008   Hong Kong   370 - 377  2008

    DOI

  • Comparative Association Rules Mining using Genetic Network Programming(GNP) with attribute Accumulation Mechanism and Its Application to Traffic Systems

    W. Wei, H. Zhou, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2008   Hong Kong   292 - 298  2008

    DOI

  • Double-Deck Elevator Systems Adaptive to Traffic Flows using Genetic Network Programming

    J. Zhou, L. Yu, S. Mabu, K. Shimada, K. Hirasawa, S. Markon

    CEC 2008   Hong Kong   773 - 778  2008

    DOI

  • Evaluating Class Association Rules using Genetic Relation Programming

    E. Gonzales, K. Taboada, K. Shimada, S. mabu, K. Hirasawa

    CEC 2008   Hong Kong   731 - 736  2008

    DOI

  • A Personalaized Association Rule Ranking Method Based on Semantic Similarity and Evolutionary Computation

    G. Yang, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2008   Hong Kong   487 - 494  2008

    DOI

  • Genetic network programming with rules

    Fengming Ye, Shigo Mabu, Kaoru Shimada, Kotaro Hirasawa

    2008 IEEE Congress on Evolutionary Computation, CEC 2008   Hong Kong   413 - 418  2008

     View Summary

    Genetic Network Programming (GNP) is an evolutionary approach which can evolve itself and find the optimal solutions. As many papers have demonstrated that GNP which has a directed graph structure can deal with dynamic environments very efficiently and effectively. It can be used in many areas such as data mining, forecasting stock markets, elevator system problems, etc. In order to improve GNP's performance further, this paper proposes a method called GNP with Rules. The aim of the proposal method is to balance exploitation and exploration, that is, to strengthen exploitation ability by using the exploited information extensively during the evolution process of GNP. The proposal method consists of 4 steps: rule extraction, rule selection, individual reconstruction and individual replacement. Tile-world was used as a simulation environment. The simulation results show some advantages of GNP with Rules over conventional GNPs. © 2008 IEEE.

    DOI

  • Time Related Association Rule Mining with Attribute Accumulation Mechanism and Its Application to Traffic Prediction

    H. Zhou, W. Wei, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2008   Hong Kong   305 - 311  2008

    DOI

  • Genetic Network Programming Based Data Mining Method for Extracting Fuzzy Association Rules

    K. Taboada, E. Gonzales, K. Shimada, S. Mabu, K. Hirasawa

    CEC 2008     1756 - 1763  2008

    DOI

  • Optimal route of road networks by dynamic programming

    Manoj Kanta Mainali, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa

    Proceedings of the International Joint Conference on Neural Networks     3416 - 3420  2008

     View Summary

    This paper introduces an iterative Q value updating algorithm based on dynamic programming for searching the optimal route and its optimal traveling time for a given Origin-Destination (OD) pair of road networks. The proposed algorithm finds the optimal route based on the local traveling time information available at each adjacent intersection. For all the intersections of the road network, Q values are introduced for determining the optimal route. When the Q values converge, we can get the optimal route from multiple sources to single destination. If there exist multiple routes with the same traveling time, the proposed method can And all of it. When the traveling time of the road links change, an alternative optimal route is found starting with the already obtained Q values. The proposed method was applied to a grid like road network and the results show that the optimal route can be found in a small number of iterations. © 2008 IEEE.

    DOI

  • Construction of Portfolio Optimization System using Genetic Network Programming with Control Nodes

    Y. Chen, S. Mabu, K. Shimada, K. Hirasawa

    GECCO 2008   Atlanta   1693 - 1694  2008

  • Stock Trading Strategies by Genetic Network Programming with Flag Nodes

    S. Mabu, Y. Chen, E. Ohkawa, K. Hirasawa

    GECCO 2008   Atlanta   1709 - 1710  2008

  • Varying Portfolio Construction of Stocks Using Genetic Network Programming with Control Nodes

    E. Ohkawa, Y. Chen, S. Mabu, K. Shimada, K. Hirasawa

    GECCO 2008   Atlanta   1715 - 1716  2008

  • double-Deck Elevator System Using Genetic Network Programming with Genetic Operators based on Pheromone Information

    L. Yu, J. Zhou, F. Ye, S. Mabu, K. Shimada, K. Hirasawa

    GECCO 2008   Atlanta   2239 - 2244  2008

  • Exceptional Association Rule Mining Using Genetic Network Programming

    K. Shimada, K. Hirasawa

    International Conference on Data Mining, DMIN 2008   Las Vegas   277 - 283  2008

  • A Nonlinear Semantic Model for Selecting Association Rules for Users

    G. Yang, K. Shimada, S. Mabu, K. Hirasawa

    SICE 2008   Tokyo   121 - 126  2008

  • Multi-car elevator system using genetic network programming

    Lu Yu, Jin Zhou, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa, Sandor Markon

    Proceedings of the SICE Annual Conference   Tokyo   127 - 131  2008

     View Summary

    Elevator group control systems are the control systems that systematically manage elevators in order to transport passengers efficiently. With the increasing need for high-performance transportation systems in buildings, multi-car elevators where two cars operate separately and independently in an elevator shaft are attracting attention as the next novel elevator system. Genetic Network, Programming(GNP) can introduce various priori knowledge of the elevator systems in its node functions easily and execute an efficient rule-based group control that is optimized evolutionary. This paper discusses the development of controllers for Multi-Car Elevator System(MCES) using GNP. The effects for MCES are examined, and we compare the advantages and performances between MCES and Double-Deck Elevator System(DDES). © 2008 SICE.

    DOI

  • Multi-objective Optimal Rout Search for Road Networks by Dynamic Programming

    M. Mainali, K. Shimada, S. Mabu, K. Hirasawa

    SICE 2008   Tokyo   628 - 632  2008

  • A stock trading model for multi-brands optimization based on genetic network programming with control nodes

    Yan Chen, Etsushi Ohkawa, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    Proceedings of the SICE Annual Conference   Tokyo   664 - 669  2008

     View Summary

    Many evolutionary computation methods applied to the financial field have been reported. A new evolutionary method named "Genetic Network Programming" (GNP) has been developed and applied to the stock market recently. The efficient trading rules created by GNP has been confirmed in our previous research. In this paper a Multi-Brands optimization system based on Genetic Network Programming with control nodes is presented. This method makes use of the information from Technical Indices and Candlestick Chart. The proposed optimization system, consisting of technical analysis rules, are trained to generate trading advice. The experimental results on the Japanese stock market show that the proposed optimization system using GNP with control nodes method outperforms other traditional models in terms of both accuracy and efficiency. © 2008 SICE.

    DOI

  • Traffic flow prediction with genetic network programming

    Wei Wei, Huiyu Zhou, Manoj Kanta Mainali, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa

    Proceedings of the SICE Annual Conference   Tokyo   670 - 675  2008

     View Summary

    In this paper, a method for traffic flow prediction has been proposed to obtain prediction rules from the past traffic data using Genetic Network Programming(GNP). GNP is an evolutionary approach which can evolve itself and find the optimal solutions. It has been clarified that GNP works well especially in dynamic environments since GNP is consisted of graphic structures, creates quite compact programs and has an implicit memory function. In this paper, GNP is applied to create a traffic flow prediction model. And we proposed the spatial adjacency model for the prediction and two kinds of models for N-step prediction. Additionally, the adaptive penalty functions are adopted for the fitness function in order to alleviate the infeasible solutions containing loops in the training process. Furthermore, the sharing function is also used to avoid the premature convergence. © 2008 SICE.

    DOI

  • Class Association Rules Mining with Time Series and Its Application to Traffic Load prediction

    H. Zhou, W. Wei, M. Mainali, K. Shimada, S. Mabu, K. Hirasawa

    SICE 2008   Tokyo   1187 - 1192  2008

  • Solving Multi-objective Optimization Problem by RasID-GA

    M. Ogata, D. Sohn, S. Mabu, K. Shimada, K. Hirasawa

    SICE 2008   Tokyo   1193 - 1198  2008

  • Discovering fuzzy classification rules using genetic network programming

    Karla Taboada, Eloy Gonzales, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa

    Proceedings of the SICE Annual Conference   Tokyo   1788 - 1793  2008

     View Summary

    Classification rule mining is an active data mining research area. Most related studies have shown how binary valued datasets are handled. However, datasets in real-world applications, usually consist of fuzzy and quantitative values. As a result, the idea to combine the different approaches with Fuzzy Set Theory has been applied more frequently in recent years. Fuzzy sets can help to overcome the so-called sharp boundary problem by allowing partial memberships to the different sets, not only 1 and 0. On the other hand, Fuzzy Sets Theory has been shown to be a very useful tool because the mined rules are expressed in linguistic terms, which are more natural and understandable for human beings. This paper proposes the combination of Fuzzy set theory and "Genetic Network Programming" (GNP) for discovering fuzzy classification rules from given quantitative data. GNP, as an extension of Genetic Algorithms (GA) and Genetic Programming (GP), is an evolutionary optimization technique that uses directed graph structures as genes instead of strings and trees
    this feature contributes creating quite compact programs and implicitly memorizing past action sequences. At last, experimental results conducted on a real world database verify the performance of the proposed method. © 2008 SICE.

    DOI

  • A Q Value-based Dynamic Programming Algorithm with Boltzmann Distribution for Optimizing the Global Traffic Routing Strategy

    S. Yu, H. Wang, F. Ye, S. Mabu, K. Shimada, K. Hirasawa

    SICE 2008   Tokyo   619 - 622  2008

  • A Global Routing Strategy in Dynamic Traffic Environments with a Combination of Q Value-based Dynamic Programming and Boltzmann Distribution

    S. Yu, F. Ye, H. Wang, S. Mabu, K. Shimada, S. Yu, K. Hirasawa

    SICE 2008   Tokyo   623 - 627  2008

    DOI

  • Evaluation of varying portfolio construction of stocks using genetic network programming with control nodes

    Etsushi Ohkawa, Yan Chen, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    Proceedings of the SICE Annual Conference   Tokyo   1231 - 1236  2008

     View Summary

    A new evolutionary method named "Genetic Network Programming with control nodes, GNPcn" has been applied to determine the timing of buying or selling stocks. GNPcn represents its solutions as directed graph structures which has some useful features inherently. For example, GNPcn has an implicit memory function which memorizes the past action sequences of agents and GNPcn can re-use nodes repeatedly in the network flow, so very compact graph structures can be made. GNPcn can determine the strategy of buying and selling stocks of multi issues. And GNPcn can distribute the purchase capital to each stock based on the distribution ratio. The effectiveness of the proposed method is confirmed by simulations. © 2008 SICE.

    DOI

  • Idle cage assignment algorithm-embedded controller of dould-deck elevator systems using genetic network programming

    Jin Zhou, Lu Yu, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa, S. Andor Markon

    Proceedings of the SICE Annual Conference   Tokyo   146 - 150  2008

     View Summary

    So far, many studies on Double-Deck Elevator Systems (DDES) have been done for exploring some more efficient algorithms to improve the system transportation capacity, especially in a heavy traffic mode. The main idea of these algorithms is to decrease the number of stops during a round trip by grouping the passengers with the same destination as much as possible. Unlike what happens in this mode, where all cages almost always keep moving, there is the case, where some cages become idle in a light traffic mode. Therefore, how to dispatch these idle cages, which is seldom considered in the heavy traffic mode, becomes important when developing the controller of DDES. In this paper, we propose a DDES controller using genetic network programming with idle cage assignment algorithm embedded for a light traffic mode. © 2008 SICE.

    DOI

  • Genetic network programming with rule chains

    Fengming Ye, Shigo Mabu, Kaoru Shimada, Kotaro Hirasawa

    Proceedings of the SICE Annual Conference   Tokyo   1220 - 1225  2008

     View Summary

    Genetic Network Programming (GNP) is a newly developed evolutionary approach which can evolve itself and And the optimal solutions. A lot of research has been done and it has been demonstrated that GNP which has a directed graph structure can deal with dynamic environments very efficiently and effectively. It can be used in many areas such data mining, elevator supervising control systems, the strategy of buying and selling stocks in stock markets, forecasting the traffic volumes in road networks, etc. In order to improve GNP's performance further, this paper proposes a method called GNP with Rule Chains. The aim of the proposed method is to balance exploitation and exploration, that is, to strengthen exploitation ability by using the exploited information extensively during the evolution process of GNP. The proposed method consists of 4 steps: rule extraction, rule selection, individual reconstruction and individual replacement. Tileworld was used as a simulation environment. The simulation results show some advantages of GNP with Rule Chains over conventional GNPs. © 2008 SICE.

    DOI

  • Buying and selling stocks of multi brands using Genetic Network Programming with control nodes

    Etsushi Ohkawa, Yan Chen, Zhiguo Bao, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   128 ( 12 ) 13 - 1819  2008

     View Summary

    A new evolutionary method named "Genetic Network Programming with control nodes, GNPcn" has been applied to determine the timing of buying or selling stocks. GNPcn represents its solutions as directed graph structures which has some useful features inherently. For example, GNPcn has an implicit memory function which memorizes the past action sequences of agents and GNPcn can re-use nodes repeatedly in the network flow, so very compact graph structures can be made. GNPcn can determine the strategy of buying and selling stocks of multi issues. The effectiveness of the proposed method is confirmed by simulations. © 2008 The Institute of Electrical Engineers of Japan.

    DOI CiNii

  • A Global Optimization Method RasID-GA for Neural Network Training

    D. Sohn, S. Mabu, K. Shimada, K. Hirasawa, J. Hu

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 1 ) 85 - 93  2008

  • A Genetic Network Programming Based Method to Mine Generalized Association Rules with Ontology

    G. Yang, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 1 ) 63 - 76  2008

  • Support Vector Machine Classifier with WHM Offset for Unbalanced Data

    B. Li, J. Hu, K. Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 1 ) 94 - 101  2008

  • Trading Rules on Stock Markets Using Genetic Network Programming with Sarsa Learning

    Y. Chen, S. Mabu, K. Shimada, K Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 4 ) 383 - 392  2008

  • Comparative Association Rules Mining Using Genetic Network Programming(GNP) with Attributes Accumulation Mechanism and its Application to Traffic systems

    W. Wei, H. Zhou, K. Shimada, S. Mabu, K. Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 4 ) 393 - 403  2008

  • Time Related Association Rules Mining with Attribute Accumulation Mechanism and Its Application to Traffic Prediction

    H. Zhou, W. Wei, K. Shimada, S. Mabu, K. Hirasawa

    Journal of Advanced Computational Intelligence and Intelligent Informatics   12 ( 5 ) 467 - 478  2008

  • Optimal Route Based on Dynamic Programming for Road networks

    M. K. Mainali, K. Shimada, S. Mabu, K. Hirasawa

    Journal of Advanced Computer Intelligence and Intelligent Informatics   12 ( 6 ) 546 - 553  2008

  • A graph-based evolutionary algorithm: Genetic Network Programming (GNP) and its extension using reinforcement learning

    Shingo Mabu, Kotaro Hirasawa, Jinglu Hu

    Evolutionary Computation   15 ( 3 ) 369 - 398  2007.09

     View Summary

    This paper proposes a graph-based evolutionary algorithm called Genetic Network Programming (GNP). Our goal is to develop GNP, which can deal with dynamic environments efficiently and effectively, based on the distinguished expression ability of the graph (network) structure. The characteristics of GNP are as follows. 1) GNP programs are composed of a number of nodes which execute simple judgment/processing, and these nodes are connected by directed links to each other. 2) The graph structure enables GNP to re-use nodes, thus the structure can be very compact. 3) The node transition of GNP is executed according to its node connections without any terminal nodes, thus the pst history of the node transition affects the current node to be used and this characteristic works as an implicit memory function. These structural characteristics are useful for dealing with dynamic environments. Furthermore, we propose an extended algorithm, "GNP with Reinforcement Learning (GNP-RL)" which combines evolution and reinforcement learning in order to create effective graph structures and obtain better results in dynamic environments. In this paper, we applied GNP to the problem of determining agents' behavior to evaluate its effectiveness. Tileworld was used as the simulation environment. The results show some advantages for GNP over conventional methods. © 2007 by the Massachusetts Institute of Technology.

    DOI PubMed CiNii

  • A graph-based evolutionary algorithm: Genetic Network Programming (GNP) and its extension using reinforcement learning

    Shingo Mabu, Kotaro Hirasawa, Jinglu Hu

    Evolutionary Computation   15 ( 3 ) 369 - 398  2007.09

     View Summary

    This paper proposes a graph-based evolutionary algorithm called Genetic Network Programming (GNP). Our goal is to develop GNP, which can deal with dynamic environments efficiently and effectively, based on the distinguished expression ability of the graph (network) structure. The characteristics of GNP are as follows. 1) GNP programs are composed of a number of nodes which execute simple judgment/processing, and these nodes are connected by directed links to each other. 2) The graph structure enables GNP to re-use nodes, thus the structure can be very compact. 3) The node transition of GNP is executed according to its node connections without any terminal nodes, thus the pst history of the node transition affects the current node to be used and this characteristic works as an implicit memory function. These structural characteristics are useful for dealing with dynamic environments. Furthermore, we propose an extended algorithm, "GNP with Reinforcement Learning (GNP-RL)" which combines evolution and reinforcement learning in order to create effective graph structures and obtain better results in dynamic environments. In this paper, we applied GNP to the problem of determining agents' behavior to evaluate its effectiveness. Tileworld was used as the simulation environment. The results show some advantages for GNP over conventional methods. © 2007 by the Massachusetts Institute of Technology.

    DOI PubMed CiNii

  • An approximate stability analysis of nonlinear systems described by Universal Learning Networks

    Kotaro Hirasawa, Shingo Mabu, Shinji Eto, Jinglu Hu

    APPLIED SOFT COMPUTING   7 ( 3 ) 642 - 651  2007.06

     View Summary

    Stability is one of the most important subjects in control systems. As for the stability of nonlinear dynamical systems, Lyapunov's direct method and linearized stability analysis method have been widely used. But, it is generally recognized that finding an appropriate Lyapunov function is fairly difficult especially for the nonlinear dynamical systems, and also it is not so easy for the linearized stability analysis to find the locally asymptotically stable region. Therefore, it is crucial and highly motivated to develop a new stability analysis method, which is easy to use and can easily study the locally asymptotically stable region at least approximately, if not exactly. On the other hand, as for the calculation of the higher order derivative, Universal Learning Networks ( ULNs) are equipped with a systematic mechanism that calculates their first and second order derivatives exactly.
    So, in this paper, an approximate stability analysis method based on h approximation is proposed in order to overcome the above problems and its application to a nonlinear dynamical control system is discussed. The proposed method studies the stability of the original trajectory by investigating whether the perturbed trajectory can approach the original trajectory or not. The above investigation is carried out approximately by using the higher order derivatives of ULNs.
    In summarizing the proposed method, firstly, the absolute values of the first order derivatives of any nodes of the trajectory with respect to any initial disturbances are calculated by using ULNs. If they approach zero at time infinity, then the trajectory is locally asymptotically stable. This is an alternative linearized stability analysis method for nonlinear trajectories without calculating Jacobians directly. In the method, the stability analysis of time-varying systems with multi-branches having any sample delays is possible, because the systems are modeled by ULNs. Secondly, the locally asymptotically stable region, where asymptotical stability is secured approximately, is obtained by finding the area where the first order terms of Taylor expansion are dominant compared to the second order terms with h approximation assuming that the higher order terms more than the third order are negligibly small in the area.
    Simulations of an inverted pendulum balancing system are carried out. From the results of the simulations, it is clarified that the stability of the inverted pendulum control system is easily analyzed by the proposed method in terms of studying the locally asymptotically stable region. (c) 2005 Elsevier B. V. All rights reserved.

    DOI

  • An approximate stability analysis of nonlinear systems described by Universal Learning Networks

    Kotaro Hirasawa, Shingo Mabu, Shinji Eto, Jinglu Hu

    APPLIED SOFT COMPUTING   7 ( 3 ) 642 - 651  2007.06

     View Summary

    Stability is one of the most important subjects in control systems. As for the stability of nonlinear dynamical systems, Lyapunov's direct method and linearized stability analysis method have been widely used. But, it is generally recognized that finding an appropriate Lyapunov function is fairly difficult especially for the nonlinear dynamical systems, and also it is not so easy for the linearized stability analysis to find the locally asymptotically stable region. Therefore, it is crucial and highly motivated to develop a new stability analysis method, which is easy to use and can easily study the locally asymptotically stable region at least approximately, if not exactly. On the other hand, as for the calculation of the higher order derivative, Universal Learning Networks ( ULNs) are equipped with a systematic mechanism that calculates their first and second order derivatives exactly.
    So, in this paper, an approximate stability analysis method based on h approximation is proposed in order to overcome the above problems and its application to a nonlinear dynamical control system is discussed. The proposed method studies the stability of the original trajectory by investigating whether the perturbed trajectory can approach the original trajectory or not. The above investigation is carried out approximately by using the higher order derivatives of ULNs.
    In summarizing the proposed method, firstly, the absolute values of the first order derivatives of any nodes of the trajectory with respect to any initial disturbances are calculated by using ULNs. If they approach zero at time infinity, then the trajectory is locally asymptotically stable. This is an alternative linearized stability analysis method for nonlinear trajectories without calculating Jacobians directly. In the method, the stability analysis of time-varying systems with multi-branches having any sample delays is possible, because the systems are modeled by ULNs. Secondly, the locally asymptotically stable region, where asymptotical stability is secured approximately, is obtained by finding the area where the first order terms of Taylor expansion are dominant compared to the second order terms with h approximation assuming that the higher order terms more than the third order are negligibly small in the area.
    Simulations of an inverted pendulum balancing system are carried out. From the results of the simulations, it is clarified that the stability of the inverted pendulum control system is easily analyzed by the proposed method in terms of studying the locally asymptotically stable region. (c) 2005 Elsevier B. V. All rights reserved.

    DOI

  • アクタークリティクを用いた遺伝的ネットワークプログラミングによる株式売買モデル

    間普真吾, 陳艶, 平澤宏太郎, 古月敬之

    システム制御情報学会研究発表講演会   京都   647 - 648  2007

    DOI

  • Time Related Association Rules Mining and Its Application to Traffic Control

    H. Zhou, W. Wei, K. Shimada, S. Mabu, K. Hirasawa

    インテリジェントシンポジウム   名古屋   97 - 102  2007

  • Comparative Association Rules Mining using Genetic Network Programming(GNP) and Its Application to Traffic Content

    W. Wei, H. Zhou, K. Shimada, S. Mabu, K. Hirasawa

    インテリジェントシンポジウム   名古屋   103 - 108  2007

  • マルチエージェントシステムにおける調整ノード付きGenetic Network Programming

    島田宗明, 江藤慎治, 間普真吾, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   名古屋   133 - 138  2007

  • 重要度指標と調整ノードを用いた機能局在型Genetic Network Programmingの構成

    江藤慎治, 間普真吾, 嶋田香, 平澤宏太郎

    インテリジェントシステムシンポジウム   名古屋   139 - 144  2007

  • フラグノード付き遺伝的ネットワークプログラミングによる株式売買モデル

    間普真吾, 平澤宏太郎, 古月敬之

    電気学会電子・情報・システム部門大会   大阪   1173 - 1179  2007

  • Search for the Optimal Traveling Time of Road Networks by Dynamic Programming

    M. Mainali, K. Shimada, S. Mabu, K. Hirasawa

    SICE 九州 2007     65 - 68  2007

  • Online Auction System with Genetic Network Programming

    N. An, K. Hirasawa, S. Mabu

    SICE 九州 2007     61 - 64  2007

  • A Hierarchical Learning System Incorporating with Supervised, Unsupervised and Reinforcement Learning

    J. Hu, T. Sasakawa, K. Hirasawa, H. Zheng

    ISNN 2007   Nanjing   403 - 411  2007

  • Trading Rules on Stock Markets using Genetic Network Programming with Sarsa Learning

    Y. Chen, S. Mabu, K. Hirasawa, J. Hu

    GECCO 2007   London   1503 - 1503  2007

    DOI

  • Effects of passenger's arrival distribution to double-deck elevator group supervisory control systems using genetic network programming

    Lu Yu, Jin Zhou, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu, Sandor Markon

    Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference   London   1476 - 1483  2007

     View Summary

    The Elevator Group Supervisory Control Systems (EGSCS) are the control systems that systematically manage three or more elevators in order to efficiently transport the passengers in buildings. Double-deck elevators, where two cages are connected with each other, are expected to be the next generation elevator systems. Meanwhile, Destination Floor Guidance Systems (DFGS) are also expected in Double-Deck Elevator Systems (DDES). With these, the passengers could be served at two consecutive floors and could input their destinations at elevator halls instead of conventional systems without DFGS. Such systems become more complex than the traditional systems and require new control methods Genetic Network Programming (GNP), a graph-based evolutionary method, has been applied to EGSCS and its advantages are shown in some previous papers. GNP can obtain the strategy of a new hall call assignment to the optimal elevator because it performs crossover and mutation operations to judgment nodes and processing nodes. In studies so far, the passenger's arrival has been assumed to take Exponential distribution for many years. In this paper, we have applied Erlang distribution and Binomial distribution in order to study how the passenger's arrival distribution affects EGSCS. We have found that the passenger's arrival distribution has great influence on EGSCS. It has been also clarified that GNP makes good performances under different conditions. Copyright 2007 ACM.

    DOI

  • Association Rule Mining for Continuous Attributes using Genetic Network Programming

    K. Toboada, K. Shimada, S. Mabu, K. Hirasawa, J Hu

    GECCO 2007   London   1578 - 1578  2007

    DOI

  • GARS: An Improved Genetic Algorithm with Reserve Selection for Global Optimization

    Y. Chen, J. Hu, K. Hirasawa, S. Yu

    GECCO 2007   London   1173 - 1178  2007

    DOI

  • Genetic Network Programming with Pararell Processing for Association Rule Mining in Large and Dense Databases

    E. Gonzales, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    GECCO 2007   London   1512 - 1512  2007

    DOI

  • Genetic Network Programming with Actor Critic and Its Application to Stock Trading Model

    S. Mabu, Y. Chen, K. Hirasawa, J. Hu

    GECCO 2007   London   2263 - 2263  2007

    DOI

  • Enhancement of Trading Rules on Stock Markets Using Genetic Network Programming with Sarsa Learning

    Y. Chen, S. Mabu, K. Hirasawa, J. Hu

    SICE 2007   kagawa   2700 - 2007  2007

  • genetic Network Programming with Class Association Rule Acquisition Mechanisms from Incomplete Databases

    K. Shimada, K. Hirasawa, J. Hu

    SICE 2007   Kagawa   2708 - 2714  2007

  • A Genetic Network Programming based method to Mine Generalized Association Rules with Ontology

    G. Yang, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    SICE 2007   Kagawa   2715 - 2722  2007

  • Association Rules Mining for Handling Continuous Attributes using Genetic Network Programming and Fuzzy Membership Functions

    K. Taboada, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    SICE 2007   Kagawa   2723 - 2729  2007

  • Genetic Network Programming with Control Nodes considering Breadth and Depth

    S. Eto, S. Mabu, K. Hirasawa, T. Hurutsuki

    SICE 2007   kagawa   470 - 475  2007

  • Double-deck Elevator Systems using Genetic Network Programming based on Variance Information

    J. Zhou, L. Yu, S. Mabu, K. Hirasawa, J. Hu, S. Markon

    SICE 2007   Kagawa   163 - 169  2007

    DOI

  • A New Cooperative Approach to Discrete Particle Swarm Optimization

    Y. Xu, J. Hu, K. Hirasawa, X. Pang

    SICE 2007   Kagawa   1311 - 1318  2007

  • Elevator group control system using genetic network programming with ACO considering transitions

    Lu Yu, Jin Zhou, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu, Sandor Markon

    Proceedings of the SICE Annual Conference   Kagawa   1330 - 1336  2007

     View Summary

    Genetic Programming Network (GNP), a graph-based evolutionary method, has been proposed several years ago as an extension of Genetic Algorithm (GA) and Genetic Programming (GP). The behavior of GNP is characterized by a balance between exploitation and exploration. To improve the evolving speed and efficiency of GNP, we developed a hybrid algorithm that combines GNP with Ant Colony Optimization (ACO). Pheromone information in the algorithm is updated not only by the fitness but also the frequency of the transitions as dynamic updating. We applied the hybrid algorithm to Elevator Group Supervisory Control Systems (EGSCS), a complex real-world problem. Finally, the simulations verified the efficacy of our proposed method. © 2007 SICE.

    DOI

  • Optimizing Reserve Size in Genetic Algorithms with Reserve Selection Using Reinforcement Learning

    Y. Chen, J. Hu, K. Hirasawa, S. Yu

    SICE 2007     1341 - 1347  2007

  • Buying and Selling of Multi Brands using Genetic Network Programming with Control Nodes

    Z. Bao, S. Mabu, K. Hirasawa, J. Hu

    SICE 2007   Kagawa   1569 - 1576  2007

    DOI

  • Hierarchical association rule mining in large and dense databases using genetic network programming

    Eloy Gonzales, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu

    Proceedings of the SICE Annual Conference   Kagawa   2686 - 2693  2007

     View Summary

    In this paper we propose a new hierarchical method to extract association rules from large and dense datasets using Genetic Network Programming (GNP) considering a real world database with a huge number of attributes. It uses three ideas. First, the large database is divided into many small datasets. Second, these small datasets are independently processed by the conventional GNP-based mining method (CGNP) in parallel. This level of processing is called Local Level. Finally, new genetic operations are carried out for small datasets considered as individuals in order to improve the number of rules extracted and their quality as well. This level of processing is called Global Level. The amount of small datasets is also important especially for avoiding the overload and improving the general performance
    we find the minimum amount of files needed to extract important association rules. The proposed method shows its effectiveness in simulations using a real world large and dense database. © 2007 SICE.

    DOI

  • Class Aoosociation Mining for Large and Dense Database with Parallel Processing of Genetic Network Programming

    E. Gonzales, K. Taboada, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    IEEE CEC 2007   Singapore   4615 - 4622  2007

    DOI

  • Double-deck elevator group supervisory control system using genetic network programming with ant colony optimization

    Lu Yu, Jin Zhou, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu, Sandor Markon

    2007 IEEE Congress on Evolutionary Computation, CEC 2007   Singapore   1015 - 1022  2007

     View Summary

    Recently, Artificial Intelligence (AI) technology has been applied to many applications. As an extension of Genetic Algorithm (GA) and Genetic Programming (GP), Genetic Network Programming (GNP) has been proposed, whose gene is constructed by directed graphs. GNP can perform a global searching, but its evolving speed is not so high and its optimal solution is hard to obtain in some cases because of the lack of the exploitation ability of it. To alleviate this difficulty, we developed a hybrid algorithm that combines Genetic Network Programming (GNP) with Ant Colony Optimization (ACO). Our goal is to introduce more exploitation mechanism into GNP. In this paper, we applied the proposed hybrid algorithm to a complicated real world problem, that is, Elevator Group Supervisory Control System (EGSCS). The simulation results showed the effectiveness of the proposed algorithm. © 2007 IEEE.

    DOI

  • Genetic network programming with control nodes

    Shinji Eto, Shingo Mabu, Kotaro Hirasawa, Takayuki Huruzuki

    2007 IEEE Congress on Evolutionary Computation, CEC 2007   Singapore   1023 - 1028  2007

     View Summary

    Many methods of generating behavior sequences of agents by evolution have been reported. A new evolutionary computation method named Genetic Network Programming (GNP) has been also developed recently along with these trends. GNP has a directed graph structure and the search for obtaining optimal GNP becomes difficult when the scale of GNP is large. The aim of this paper is to find a well structured GNP considering Breadth and Depth of GNP searching. It has been shown that the proposed method is efficient compared with conventional GNPs from simulations using a garbage collector problem. © 2007 IEEE.

    DOI

  • Mining Association Rules from Databases with Continuous Attributes using Genetic Network Programming

    K. Toboada, E. Gonzales, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    IEEE CEC 2007   Singapore   1311 - 1317  2007

    DOI

  • Double-deck elevator systems using genetic network programming with reinforcement learning

    Jin Zhou, Lu Yu, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu, Sandor Markon

    2007 IEEE Congress on Evolutionary Computation, CEC 2007   Singapore   2025 - 2031  2007

     View Summary

    In order to increase the transportation capability of elevator group systems in high-rise buildings without adding elevator installation space, double-deck elevator system (DDES) is developed as one of the next generation elevator group systems. Artificial intelligence (AI) technologies have been employed to And some efficient solutions in the elevator group control systems during the late 20th century. Genetic Network Programming (GNP), a new evolutionary computation method, is reported to be employed as the elevator group system controller in some studies of recent years. Moreover, reinforcement learning (RL) is also verified to be useful for more improvements of elevator group performances when it is combined with GNP. In this paper, we proposed a new approach of DDES using GNP with RL, and did some experiments on a simulated elevator group system of a typical office building to check its efficiency. Simulation results show that the DDES using GNP with RL performs better than the one without RL in regular and down-peak time, while both of them outperforms a conventional approach and a heuristic approach in all three traffic patterns. ©2007 IEEE.

    DOI

  • Performance Tuning of Genetic Algorithms with Reserve Selection

    Y. Chen, J. Hu, K. Hirasawa, S. Yu

    IEEE CEC 2007   Singapore   2202 - 2209  2007

    DOI

  • Mining Equalized Association Rules from Multi Concept Layers of Ontology Using Genetic Network Programming

    G. Yang, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    IEEE CEC 2007   Singapore   705 - 712  2007

    DOI

  • Training of multi-branch neural networks using RasID-GA

    Dongkyu Sohn, Shingo Mabu, Kaoru Shimada, Kotaro Hirasawa, Jingiu Hu

    2007 IEEE Congress on Evolutionary Computation, CEC 2007   Singapore   2064 - 2070  2007

     View Summary

    This paper applies a Adaptive Random search with Intensification and Diversification combined with Genetic Algorithm (RasID-GA) to neural network training. In the previous work, we proposed RasID-GA which combines the best properties of RasID and Genetic Algorithm for optimization. Neural networks are widely used in pattern recognition, system modeling, prediction and other areas. Although most neural network training uses gradient based schemes such as well-known back-propagation (BP), but sometimes BP is easily dropped into local minima. In this paper, we train multi-branch neural networks using RasID-GA with constraint coefficient C by which the feasible solution space is controlled. In addition, we use Mackey-Glass time prediction to test a generalization ability of the proposed method. ©2007 IEEE.

    DOI

  • Genetic Network Programming with Darsa Learning and Its Application to Creating Stock Trading Rules

    Y. Chen, S. Mabu, K. Hirasawa, J. Hu

    IEEE CEC 2007   Singapore   220 - 227  2007

    DOI

  • Stock Trading Rules Using Genetic Network Programming with Actor-Crotic

    S. Mabu, Y. Chen, K. Hirasawa, J. Hu

    IEEE CEC 2007   Singapore   508 - 515  2007

    DOI

  • Trading Rules on Stock Markets Using Genetic Network Programming with Candle Chart

    MABU Shingo, IZUMI Yoshihiro, HIRASAWA Kotaro, FURUZUKI Takayuki

    Transactions of the Society of Instrument and Control Engineers   43 ( 4 ) 317 - 322  2007

     View Summary

    A new evolutionary method named 'Genetic Network Programming, GNP' has been proposed. GNP represents its solutions as directed graph structures which have some useful features inherently. For example, GNP has the implicit memory function which memorizes the past action sequences of agents, and GNP can re-use nodes repeatedly in the network flow, so very compact graph structures can be made. In this paper, the stock trading model using GNP with Candle Chart is proposed and its effectiveness is comfirmed by trading simulations.

    DOI CiNii

  • Trading Rules on Stock Markets Using Genetic Network Programming with Reinforcement Learning and Importance Index

    MABU Shingo, HIRASAWA Kotaro, FURUZUKI Takayuki

    IEEJ Transactions on Electronics, Information and Systems   127 ( 7 ) 1061 - 1067  2007

     View Summary

    Genetic Network Programming (GNP) is an evolutionary computation which represents its solutions using graph structures. Since GNP can create quite compact programs and has an implicit memory function, it has been clarified that GNP works well especially in dynamic environments. In addition, a study on creating trading rules on stock markets using GNP with Importance Index (GNP-IMX) has been done. IMX is a new element which is a criterion for decision making. In this paper, we combined GNP-IMX with Actor-Critic (GNP-IMX&AC) and create trading rules on stock markets. Evolution-based methods evolve their programs after enough period of time because they must calculate fitness values, however reinforcement learning can change programs during the period, therefore the trading rules can be created efficiently. In the simulation, the proposed method is trained using the stock prices of 10 brands in 2002 and 2003. Then the generalization ability is tested using the stock prices in 2004. The simulation results show that the proposed method can obtain larger profits than GNP-IMX without AC and Buy&Hold.

    DOI CiNii

  • A Double-Deck Elevator Group Supervisory Control System with Destination Floor Guidance System Using Genetic Network Programming

    YU Lu, ZHOU Jin, MABU Shingo, HIRASAWA Kotaro, HU Jinglu, MARKON Sandor

    IEEJ Transactions on Electronics, Information and Systems   127 ( 7 ) 1115 - 1122  2007

     View Summary

    The Elevator Group Supervisory Control Systems (EGSCS) are the control systems that systematically manage three or more elevators in order to efficiently transport the passengers in buildings. Double-deck elevators, where two elevators are connected with each other, serve passengers at two consecutive floors simultaneously. Double-deck Elevator systems (DDES) become more complex in their behavior than conventional single-deck elevator systems (SDES). Recently, Artificial Intelligence (AI) technology has been used in such complex systems. Genetic Network Programming (GNP), a graph-based evolutionary method, has been applied to EGSCS and its advantages are shown in some papers. GNP can obtain the strategy of a new hall call assignment to the optimal elevator when it performs crossover and mutation operations to judgment nodes and processing nodes. Meanwhile, Destination Floor Guidance System (DFGS) is installed in DDES, so that passengers can also input their destinations at elevator halls. In this paper, we have applied GNP to DDES and compared DFGS with normal systems. The waiting time and traveling time of DFGS are all improved because of getting more information from DFGS. The simulations showed the effectiveness of the double-deck elevators with DFGS in different building traffics.

    DOI CiNii

  • Effects of Passengers' Arrival Distribution to Double-deck Elevator Group Supervisory Control Systems Using Genetic Network Programming

    L. Yu, J. Zhou, S. Mabu, K. Hirasawa, J. Hu, S. Markon

    計測自動制御学会産業論文集   6 ( 11 ) 85 - 92  2007

  • Elevator Group Supervisory Control System Using Genetic Network Programming with Macro Nodes and Reinforcement Learning

    ZHOU Jin, YU Lu, MABU Shingo, HIRASAWA Kotaro, HU Jinglu, MARKON Sandor

    IEEJ Transactions on Electronics, Information and Systems   127 ( 8 ) 1234 - 1242  2007

     View Summary

    Elevator Group Supervisory Control System (EGSCS) is a very large scale stochastic dynamic optimization problem. Due to its vast state space, significant uncertainty and numerous resource constraints such as finite car capacities and registered hall/car calls, it is hard to manage EGSCS using conventional control methods. Recently, many solutions for EGSCS using Artificial Intelligence (AI) technologies have been reported. Genetic Network Programming (GNP), which is proposed as a new evolutionary computation method several years ago, is also proved to be efficient when applied to EGSCS problem. In this paper, we propose an extended algorithm for EGSCS by introducing Reinforcement Learning (RL) into GNP framework, and an improvement of the EGSCS' performances is expected since the efficiency of GNP with RL has been clarified in some other studies like tile-world problem. Simulation tests using traffic flows in a typical office building have been made, and the results show an actual improvement of the EGSCS' performances comparing to the algorithms using original GNP and conventional control methods. Furthermore, as a further study, an importance weight optimization algorithm is employed based on GNP with RL and its efficiency is also verified with the better performances.

    DOI CiNii

  • マルチブランチ構造を有するリカレントニューラルネットワーク

    山下貴志, 間普真吾, 平澤宏太郎, 古月敬之

    電気学会論文誌 C   127 ( 9 ) 1430 - 1435  2007

    DOI CiNii

  • Genetic Network Programming with Actor-Critic

    H. Hatakeyama, S. Mabu, K. Hirasawa, J. Hu

    Jouranal of Adavanced Computational Intelligence and Intelligent Informatics   11 ( 1 ) 79 - 86  2007

  • Adaptation and Self Adaptation Mechanism in Genetic Network Programming for Mining Association Rules

    K. Taboada, E. Gonzales, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    Journal of Advanced Computaional Intelligence and Intelligent Informatics   11 ( 3 ) 343 - 353  2007

  • Optimization Method RasID-GA for Numerical Constrained Optimization Problems

    D. Sohn, S. Mabu, K. Hirasawa, J. Hu

    Journal of Advanced Computational Intelligence and Intelligent Informatics   11 ( 5 ) 469 - 477  2007

  • Double-Deck Elevator Group Supervisory Control System Using Genetic Network Programming with Ant Colony Optimization with Evaporation

    L. Yu, J. Zhou, S. Mabu, K. Hirasawa, J. Hu

    Journal od Advanced Computational Intelligence anf Intelligent Informatics   11 ( 9 ) 1149 - 1158  2007

  • A Novel Taxi Dispatch System Integrating a Multiple-Customer Strategy and Genetic network Programming

    Q. Meng, S. Mabu, L. Yu, K. Hirasawa

    JACIII   14 ( 5 ) 442 - 452  2007

  • Time Related Association Rules Mining and Its Application to Traffic Control

    H. Zhou, W. Wei, K. Shimada, S. Mabu, K. Hirasawa

    インテリジェントシンポジウム   名古屋   97 - 102  2007

  • Comparative Association Rules Mining using Genetic Network Programming(GNP) and Its Application to Traffic Content

    W. Wei, H. Zhou, K. Shimada, S. Mabu, K. Hirasawa

    インテリジェントシンポジウム   名古屋   103 - 108  2007

  • Search for the Optimal Traveling Time of Road Networks by Dynamic Programming

    M. Mainali, K. Shimada, S. Mabu, K. Hirasawa

    SICE 九州 2007     65 - 68  2007

  • Online Auction System with Genetic Network Programming

    N. An, K. Hirasawa, S. Mabu

    SICE 九州 2007     61 - 64  2007

  • A Hierarchical Learning System Incorporating with Supervised, Unsupervised and Reinforcement Learning

    J. Hu, T. Sasakawa, K. Hirasawa, H. Zheng

    ISNN 2007   Nanjing   403 - 411  2007

  • Trading Rules on Stock Markets using Genetic Network Programming with Sarsa Learning

    Y. Chen, S. Mabu, K. Hirasawa, J. Hu

    GECCO 2007   London   1503 - 1503  2007

    DOI

  • Association Rule Mining for Continuous Attributes using Genetic Network Programming

    K. Toboada, K. Shimada, S. Mabu, K. Hirasawa, J Hu

    GECCO 2007   London   1578 - 1578  2007

    DOI

  • GARS: An Improved Genetic Algorithm with Reserve Selection for Global Optimization

    Y. Chen, J. Hu, K. Hirasawa, S. Yu

    GECCO 2007   London   1173 - 1178  2007

    DOI

  • Genetic Network Programming with Pararell Processing for Association Rule Mining in Large and Dense Databases

    E. Gonzales, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    GECCO 2007   London   1512 - 1512  2007

    DOI

  • Genetic network programming with actor-critic and its application to stock trading model

    Shingo Mabu, Yan Chen, Kotaro Hirasawa, Jinglu Hu

    Proceedings of GECCO 2007: Genetic and Evolutionary Computation Conference   London   2263 - 2263  2007

    DOI

  • Enhancement of Trading Rules on Stock Markets Using Genetic Network Programming with Sarsa Learning

    Y. Chen, S. Mabu, K. Hirasawa, J. Hu

    SICE 2007   kagawa   2700 - 2007  2007

  • genetic Network Programming with Class Association Rule Acquisition Mechanisms from Incomplete Databases

    K. Shimada, K. Hirasawa, J. Hu

    SICE 2007   Kagawa   2708 - 2714  2007

  • A Genetic Network Programming based method to Mine Generalized Association Rules with Ontology

    G. Yang, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    SICE 2007   Kagawa   2715 - 2722  2007

  • Association Rules Mining for Handling Continuous Attributes using Genetic Network Programming and Fuzzy Membership Functions

    K. Taboada, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    SICE 2007   Kagawa   2723 - 2729  2007

  • Genetic Network Programming with Control Nodes considering Breadth and Depth

    S. Eto, S. Mabu, K. Hirasawa, T. Hurutsuki

    SICE 2007   kagawa   470 - 475  2007

  • Double-deck elevator systems using genetic network programming based on variance information

    Jin Zhou, Lu Yu, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu, Sandor Markon

    Proceedings of the SICE Annual Conference   Kagawa   163 - 169  2007

     View Summary

    Double-Deck Elevator Systems (DDES) have been invented to improve the transportation capacity of elevator group systems for decades. There are several specific features in DDES due to its specific structure, i.e., two decks are vertically connected in one shaft. Even though the DDES could work well in a pure up-peak traffic pattern by cutting up to half of the stops in an elevator round trip, it becomes intractable because of the features when running in some other traffic patterns. Some solutions employing evolutionary computation methods such as genetic algorithm were also proposed in recent years. In this paper, we propose an approach of DDES using genetic network programming based on our past studies in this field. © 2007 SICE.

    DOI

  • A New Cooperative Approach to Discrete Particle Swarm Optimization

    Y. Xu, J. Hu, K. Hirasawa, X. Pang

    SICE 2007   Kagawa   1311 - 1318  2007

  • Elevator group control system using genetic network programming with ACO considering transitions

    Lu Yu, Jin Zhou, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu, Sandor Markon

    Proceedings of the SICE Annual Conference   Kagawa   1330 - 1336  2007

     View Summary

    Genetic Programming Network (GNP), a graph-based evolutionary method, has been proposed several years ago as an extension of Genetic Algorithm (GA) and Genetic Programming (GP). The behavior of GNP is characterized by a balance between exploitation and exploration. To improve the evolving speed and efficiency of GNP, we developed a hybrid algorithm that combines GNP with Ant Colony Optimization (ACO). Pheromone information in the algorithm is updated not only by the fitness but also the frequency of the transitions as dynamic updating. We applied the hybrid algorithm to Elevator Group Supervisory Control Systems (EGSCS), a complex real-world problem. Finally, the simulations verified the efficacy of our proposed method. © 2007 SICE.

    DOI

  • Optimizing Reserve Size in Genetic Algorithms with Reserve Selection Using Reinforcement Learning

    Y. Chen, J. Hu, K. Hirasawa, S. Yu

    SICE 2007     1341 - 1347  2007

  • Buying and Selling of Multi Brands using Genetic Network Programming with Control Nodes

    Z. Bao, S. Mabu, K. Hirasawa, J. Hu

    SICE 2007   Kagawa   1569 - 1576  2007

    DOI

  • Hierarchical association rule mining in large and dense databases using genetic network programming

    Eloy Gonzales, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu

    Proceedings of the SICE Annual Conference   Kagawa   2686 - 2693  2007

     View Summary

    In this paper we propose a new hierarchical method to extract association rules from large and dense datasets using Genetic Network Programming (GNP) considering a real world database with a huge number of attributes. It uses three ideas. First, the large database is divided into many small datasets. Second, these small datasets are independently processed by the conventional GNP-based mining method (CGNP) in parallel. This level of processing is called Local Level. Finally, new genetic operations are carried out for small datasets considered as individuals in order to improve the number of rules extracted and their quality as well. This level of processing is called Global Level. The amount of small datasets is also important especially for avoiding the overload and improving the general performance
    we find the minimum amount of files needed to extract important association rules. The proposed method shows its effectiveness in simulations using a real world large and dense database. © 2007 SICE.

    DOI

  • Class association rule mining for large and dense databases with parallel processing of genetic network programming

    Eloy Gonzales, Karla Taboada, Kaoru Shimada, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu

    2007 IEEE Congress on Evolutionary Computation, CEC 2007   Singapore   4615 - 4622  2007

     View Summary

    Among several methods of extracting association rules that have been reported, a new evolutionary computation method named Genetic Network Programming (GNP) has also shown its effectiveness for small datasets that have a relatively small number of attributes. The aim of this paper is to propose a new method to extract association rules from large and dense datasets with a huge amount of attributes using GNP. It consists of two-level of processing. Server Level where conventional GNP based mining method runs in parallel and Client Level where files are considered as individuals and genetic operations are carried out over them. The algorithm starts dividing the large dataset into small datasets with appropiate size, and then each of them are dealt with GNP in parallel processing. The new association rules obtained in each generation are stored in a general global pool. We compared several genetic operators applied to the individuals in the Global Level. The proposed method showed remarkable improvements on simulations. © 2007 IEEE.

    DOI

  • Double-deck elevator group supervisory control system using genetic network programming with ant colony optimization

    Lu Yu, Jin Zhou, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu, Sandor Markon

    2007 IEEE Congress on Evolutionary Computation, CEC 2007   Singapore   1015 - 1022  2007

     View Summary

    Recently, Artificial Intelligence (AI) technology has been applied to many applications. As an extension of Genetic Algorithm (GA) and Genetic Programming (GP), Genetic Network Programming (GNP) has been proposed, whose gene is constructed by directed graphs. GNP can perform a global searching, but its evolving speed is not so high and its optimal solution is hard to obtain in some cases because of the lack of the exploitation ability of it. To alleviate this difficulty, we developed a hybrid algorithm that combines Genetic Network Programming (GNP) with Ant Colony Optimization (ACO). Our goal is to introduce more exploitation mechanism into GNP. In this paper, we applied the proposed hybrid algorithm to a complicated real world problem, that is, Elevator Group Supervisory Control System (EGSCS). The simulation results showed the effectiveness of the proposed algorithm. © 2007 IEEE.

    DOI

  • Genetic network programming with control nodes

    Shinji Eto, Shingo Mabu, Kotaro Hirasawa, Takayuki Huruzuki

    2007 IEEE Congress on Evolutionary Computation, CEC 2007   Singapore   1023 - 1028  2007

     View Summary

    Many methods of generating behavior sequences of agents by evolution have been reported. A new evolutionary computation method named Genetic Network Programming (GNP) has been also developed recently along with these trends. GNP has a directed graph structure and the search for obtaining optimal GNP becomes difficult when the scale of GNP is large. The aim of this paper is to find a well structured GNP considering Breadth and Depth of GNP searching. It has been shown that the proposed method is efficient compared with conventional GNPs from simulations using a garbage collector problem. © 2007 IEEE.

    DOI

  • Mining Association Rules from Databases with Continuous Attributes using Genetic Network Programming

    K. Toboada, E. Gonzales, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    IEEE CEC 2007   Singapore   1311 - 1317  2007

    DOI

  • Double-deck elevator systems using genetic network programming with reinforcement learning

    Jin Zhou, Lu Yu, Shingo Mabu, Kotaro Hirasawa, Jinglu Hu, Sandor Markon

    2007 IEEE Congress on Evolutionary Computation, CEC 2007   Singapore   2025 - 2031  2007

     View Summary

    In order to increase the transportation capability of elevator group systems in high-rise buildings without adding elevator installation space, double-deck elevator system (DDES) is developed as one of the next generation elevator group systems. Artificial intelligence (AI) technologies have been employed to And some efficient solutions in the elevator group control systems during the late 20th century. Genetic Network Programming (GNP), a new evolutionary computation method, is reported to be employed as the elevator group system controller in some studies of recent years. Moreover, reinforcement learning (RL) is also verified to be useful for more improvements of elevator group performances when it is combined with GNP. In this paper, we proposed a new approach of DDES using GNP with RL, and did some experiments on a simulated elevator group system of a typical office building to check its efficiency. Simulation results show that the DDES using GNP with RL performs better than the one without RL in regular and down-peak time, while both of them outperforms a conventional approach and a heuristic approach in all three traffic patterns. ©2007 IEEE.

    DOI

  • Performance Tuning of Genetic Algorithms with Reserve Selection

    Y. Chen, J. Hu, K. Hirasawa, S. Yu

    IEEE CEC 2007   Singapore   2202 - 2209  2007

    DOI

  • Mining Equalized Association Rules from Multi Concept Layers of Ontology Using Genetic Network Programming

    G. Yang, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    IEEE CEC 2007   Singapore   705 - 712  2007

    DOI

  • Training of Multi-Branch Neural Networks uisng RasID-GA

    D. Sohn, S. Mabu, K. Shimada, K. Hirasawa, J. Hu

    IEEE CEC 2007   Singapore   2064 - 2070  2007

    DOI

  • Genetic Network Programming with Darsa Learning and Its Application to Creating Stock Trading Rules

    Y. Chen, S. Mabu, K. Hirasawa, J. Hu

    IEEE CEC 2007   Singapore   220 - 227  2007

    DOI

  • Stock trading rules using genetic network programming with actor-critic

    Shingo Mabu, Yan Chen, Kotaro Hirasawa, Jinglu Hu

    2007 IEEE Congress on Evolutionary Computation, CEC 2007   Singapore   508 - 515  2007

     View Summary

    Genetic Network Programming (GNP) is an evolutionary computation which represents its solutions using graph structures. Since GNP can create quite compact programs and has an implicit memory function, it has been clarified that GNP works well especially in dynamic environments. In this paper, GNP is applied to creating a stock trading model. The first important point is to combine GNP with Actor-Critic which is one of the reinforcement learning algorithms. Evolution-based methods evolve their programs after task execution because they must calculate fitness values, while reinforcement learning can change programs during task execution, therefore the programs can be created efficiently. The second important point is that GNP with Actor-Critic (GNP-AC) can select appropriate technical indexes to judge the buying and selling timing of stocks using Importance Index especially designed for stock trading decision making. In the simulations, the trading model is trained using the stock prices of 20 brands in 2001, 2002 and 2003. Then the generalization ability is tested using the stock prices in 2004. From the simulation results, it is clarified that the trading rules of GNP-AC obtain higher profits than Buy&amp
    Hold method. © 2007 IEEE.

    DOI

  • A Double-Deck Elevator Group Supervisory Control System with Destination Floor Guidance System using Genetic Network Programming

    L. Yu, J. Zhou, S. Mabu, K. Hirasawa, J. Hu, S. Markon

    電気学会論文誌 C   127 ( 7 ) 1115 - 1122  2007

     View Summary

    The Elevator Group Supervisory Control Systems (EGSCS) are the control systems that systematically manage three or more elevators in order to efficiently transport the passengers in buildings. Double-deck elevators, where two elevators are connected with each other, serve passengers at two consecutive floors simultaneously. Double-deck Elevator systems (DDES) become more complex in their behavior than conventional single-deck elevator systems (SDES). Recently, Artificial Intelligence (AI) technology has been used in such complex systems. Genetic Network Programming (GNP), a graph-based evolutionary method, has been applied to EGSCS and its advantages are shown in some papers. GNP can obtain the strategy of a new hall call assignment to the optimal elevator when it performs crossover and mutation operations to judgment nodes and processing nodes. Meanwhile, Destination Floor Guidance System (DFGS) is installed in DDES, so that passengers can also input their destinations at elevator halls. In this paper, we have applied GNP to DDES and compared DFGS with normal systems. The waiting time and traveling time of DFGS are all improved because of getting more information from DFGS. The simulations showed the effectiveness of the double-deck elevators with DFGS in different building traffics.

    DOI CiNii

  • Effects of Passengers' Arrival Distribution to Double-deck Elevator Group Supervisory Control Systems Using Genetic Network Programming

    L. Yu, J. Zhou, S. Mabu, K. Hirasawa, J. Hu, S. Markon

    計測自動制御学会産業論文集   6 ( 11 ) 85 - 92  2007

  • Elevator Group Supervisory Control System Using Genetic Network Programming with Macro Nodes and Reinforcement Learning

    J. Zhou, L. Yu, S. Mabu, K. Hirasawa, J. Hu, S. Markon

    電気学会論文誌 C   127 ( 8 ) 1234 - 1242  2007

     View Summary

    Elevator Group Supervisory Control System (EGSCS) is a very large scale stochastic dynamic optimization problem. Due to its vast state space, significant uncertainty and numerous resource constraints such as finite car capacities and registered hall/car calls, it is hard to manage EGSCS using conventional control methods. Recently, many solutions for EGSCS using Artificial Intelligence (AI) technologies have been reported. Genetic Network Programming (GNP), which is proposed as a new evolutionary computation method several years ago, is also proved to be efficient when applied to EGSCS problem. In this paper, we propose an extended algorithm for EGSCS by introducing Reinforcement Learning (RL) into GNP framework, and an improvement of the EGSCS' performances is expected since the efficiency of GNP with RL has been clarified in some other studies like tile-world problem. Simulation tests using traffic flows in a typical office building have been made, and the results show an actual improvement of the EGSCS' performances comparing to the algorithms using original GNP and conventional control methods. Furthermore, as a further study, an importance weight optimization algorithm is employed based on GNP with RL and its efficiency is also verified with the better performances.

    DOI CiNii

  • Genetic Network Programming with Actor-Critic

    H. Hatakeyama, S. Mabu, K. Hirasawa, J. Hu

    Jouranal of Adavanced Computational Intelligence and Intelligent Informatics   11 ( 1 ) 79 - 86  2007

  • Adaptation and Self Adaptation Mechanism in Genetic Network Programming for Mining Association Rules

    K. Taboada, E. Gonzales, K. Shimada, S. Mabu, K. Hirasawa, J. Hu

    Journal of Advanced Computaional Intelligence and Intelligent Informatics   11 ( 3 ) 343 - 353  2007

  • Optimization Method RasID-GA for Numerical Constrained Optimization Problems

    D. Sohn, S. Mabu, K. Hirasawa, J. Hu

    Journal of Advanced Computational Intelligence and Intelligent Informatics   11 ( 5 ) 469 - 477  2007

  • Double-Deck Elevator Group Supervisory Control System Using Genetic Network Programming with Ant Colony Optimization with Evaporation

    L. Yu, J. Zhou, S. Mabu, K. Hirasawa, J. Hu

    Journal od Advanced Computational Intelligence anf Intelligent Informatics   11 ( 9 ) 1149 - 1158  2007

  • A Novel Taxi Dispatch System Integrating a Multiple-Customer Strategy and Genetic network Programming

    Q. Meng, S. Mabu, L. Yu, K. Hirasawa

    JACIII   14 ( 5 ) 442 - 452  2007

  • Propagation and control of stochastic signals through universal learning networks

    Kotaro Hirasawa, Shingo Mabu, Jinglu Hu

    Neural Networks   19 ( 4 ) 487 - 499  2006.05

     View Summary

    The way of propagating and control of stochastic signals through Universal Learning Networks (ULNs) and its applications are proposed. ULNs have been already developed to form a superset of neural networks and have been applied as a universal framework for modeling and control of non-linear large-scale complex systems. However, the ULNs cannot deal with stochastic variables. Deterministic signals can be propagated through a ULN, but the ULN does not provide any stochastic characteristics of the signals propagating through it. The proposed method named Probabilistic Universal Learning Networks (PrULNs) can process stochastic variables and can train network parameters so that the signals behave with the pre-specified stochastic properties. As examples of applications of the proposed method, control and identification of non-linear dynamic systems with noises are studied, and it is shown that the method are useful for dealing with the control and identification of the non-linear stochastic systems contaminated with noises. © 2005 Elsevier Ltd. All rights reserved.

    DOI PubMed CiNii

  • Propagation and control of stochastic signals through universal learning networks

    Kotaro Hirasawa, Shingo Mabu, Jinglu Hu

    NEURAL NETWORKS   19 ( 4 ) 487 - 499  2006.05

     View Summary

    The way of propagating and control of stochastic signals through Universal Learning Networks (ULNs) and its applications are proposed. ULNs have been already developed to form a superset of neural networks and have been applied as a universal framework for modeling and control of non-linear large-scale complex systems.
    However, the ULNs cannot deal with stochastic variables. Deterministic signals can be propagated through a ULN, but the ULN does not provide any stochastic characteristics of the signals propagating through it.
    The proposed method named Probabilistic Universal Learning Networks (PrULNs) can process stochastic variables and can train network parameters so that the signals behave with the pre-specified stochastic properties.
    As examples of applications of the proposed method, control and identification of non-linear dynamic systems with noises are studied, and it is shown that the method are useful for dealing with the control and identification of the non-linear stochastic systems contaminated with noises. (c) 2005 Elsevier Ltd. All rights reserved.

    DOI PubMed CiNii

  • A study of evolutionary multiagent models based on symbiosis

    T Eguchi, K Hirasawa, JL Hu, N Ota

    IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART B-CYBERNETICS   36 ( 1 ) 179 - 193  2006.02

     View Summary

    Multiagent Systems with Symbiotic Learning and Evolution (Masbiole) has been proposed and studied, which is a new methodology of Multiagent Systems (MAS) based on symbiosis in the ecosystem. Masbiole employs a method of symbiotic learning and evolution where agents can learn or evolve according to their symbiotic relations toward others, i.e., considering the benefits/losses of both itself and an opponent. As a result, Masbiole can escape from Nash Equilibria and obtain better performances than conventional MAS where agents consider only their own benefits. This paper focuses on the evolutionary model of Masbiole, and its characteristics are examined especially with an emphasis on the behaviors of agents obtained by symbiotic evolution. In the simulations, two ideas suitable for the effective analysis of such behaviors are introduced; "Match Type Tile-world (MTT)" and "Genetic Network Programming (GNP)". MTT is a virtual model where tile-world is improved so that agents can behave considering their symbiotic relations. GNP is a newly developed evolutionary computation which has the directed graph type gene structure and enables to analyze the decision making mechanism of agents easily. Simulation results show that Masbiole can obtain various kinds of behaviors and better performances than conventional MAS in MTT by evolution.

    DOI PubMed CiNii

  • A study of evolutionary multiagent models based on symbiosis

    T Eguchi, K Hirasawa, JL Hu, N Ota

    IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART B-CYBERNETICS   36 ( 1 ) 179 - 193  2006.02

     View Summary

    Multiagent Systems with Symbiotic Learning and Evolution (Masbiole) has been proposed and studied, which is a new methodology of Multiagent Systems (MAS) based on symbiosis in the ecosystem. Masbiole employs a method of symbiotic learning and evolution where agents can learn or evolve according to their symbiotic relations toward others, i.e., considering the benefits/losses of both itself and an opponent. As a result, Masbiole can escape from Nash Equilibria and obtain better performances than conventional MAS where agents consider only their own benefits. This paper focuses on the evolutionary model of Masbiole, and its characteristics are examined especially with an emphasis on the behaviors of agents obtained by symbiotic evolution. In the simulations, two ideas suitable for the effective analysis of such behaviors are introduced; "Match Type Tile-world (MTT)" and "Genetic Network Programming (GNP)". MTT is a virtual model where tile-world is improved so that agents can behave considering their symbiotic relations. GNP is a newly developed evolutionary computation which has the directed graph type gene structure and enables to analyze the decision making mechanism of agents easily. Simulation results show that Masbiole can obtain various kinds of behaviors and better performances than conventional MAS in MTT by evolution.

    DOI PubMed CiNii

  • Genetic Network Programming によるエレベータ群管理の最適化

    江口徹, 平澤宏太郎, マルコンシャンドル

    日本機械学会技術講演会論文集   東京   17 - 20  2006

  • 遺伝的ネットワークプログラミングを用いた行き先階登録方式エレベータ群管理システム

    久保太一, 江口徹, 間普真吾, 平澤宏太郎, 古月敬之, マルコン シャンドル

    情報処理学会 火の国情報シンポジウム   熊本  2006

  • 強化学習と重要度指標を用いた遺伝的ネットワークプログラミングによる株式売買モデル

    間普真吾, 高橋好史, 平澤宏太郎, 古月敬之

    電気学会C部門大会   横浜  2006

  • 多スタートノードGNPを用いた株式売買ポートフォリオの基礎検討

    白石憲正, 間普真吾, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   柏   145 - 150  2006

  • Benchmark Test of RasID-GA for Inequality/Equality Constrained Optimization

    D. Sohn, S. Mabu, K. Hirasawa, J. Hu

    インテリジェントシステムシンポジウム   柏   155 - 160  2006

  • 多エージェント間の共生進化アルゴリズム

    田中大介, 間普真吾, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   柏   187 - 192  2006

  • 遺伝的ネットワークプログラミングによる不完全なデータベースからのクラス相関ルールの抽出

    森川英治, 嶋田香, 間普真吾, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   柏   227 - 232  2006

  • イントロン付Genetic Network Programmingの基礎検討

    畑和宏, 間普真吾, 江藤慎治, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   柏   267 - 272  2006

  • 幅と深さを考慮したGenetic Network Programmingの進化手法

    江藤慎治, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   柏   263 - 266  2006

  • Actor-Criticを用いた遺伝的ネットワークプログラミング

    畠山裕之, 間普真吾, 平澤宏太郎, 古月敬之

    インテリジントシステムシンポジウム   柏   95 - 100  2006

  • Genetic Network Programmingによるダブルデッキエレベータ群管理システム

    周金、Yu Lu, 間普真吾, 平澤宏太郎, マルコンシャンドル

    計測自動制御学会システム・情報部門学術講演会   筑波市   131 - 135  2006

  • An Improved Method for Identification of Quasi-ARMAX Model

    J. Ji, J. Hu, K. Hirasawa

    SICE九州支部学術講演会   佐賀   43 - 44  2006

  • GAによる非線形多項式モデルの二ステップ同定法

    程学飛, 古月敬之, 平澤宏太郎

    SICE九州支部学術講演会   佐賀   45 - 46  2006

  • Evolutionary Method of Genetic Network Programming Considering Breadth and Depth

    S. Eto, S. Mabu, K. Hirasawa

    GECCO 2006   Seattle  2006

  • Associate Rule Mining with Chi-Squared Test Using Alternate Genetic Network Programming

    K. Shimada, K. Hirasawa, J. Hu

    ICDM 2006   Leipzig   202 - 216  2006

    DOI

  • Genetic Network Programming with Reinforcement Learning Using Sarsa Algorithm

    S. Mabu, H. Hatakeyama, K. Hirasawa, J. Hu

    CEC 2006   Vancouver   1570 - 1575  2006

  • An Extension of Genetic Network Programming with Reinforcement Learning Using Actor-Critic

    H. Hatakeyama, S. Mabu, K. Hirasawa, J. Hu

    CEC 2006   Vancouver   5686 - 5692  2006

  • Trading Rules on the Stock Markets using Genetic Network Programming with Candlestick Chart

    Y. Izumi, T. Yamaguchi, S. Mabu, K. Hirasawa, J. Hu

    CEC 2006   Vancouver   8531 - 8536  2006

  • RasID-GA with Simplex Crossover for Optimization Problems

    D. Sohn, S. Mabu, K. Hirasawa, J. Hu

    CEC 2006   Vancouver   10378 - 10385  2006

  • A study of applying Genetic Network Programming with Reinforcement Learning to Elevator Group Supervisory Control System

    J. Zhou, T. Eguchi, S. Mabu, K. Hirasawa, J. Hu, S. Markon

    CEC 2006   Vancouver   10392 - 10398  2006

  • Effective Training Method for Functional Localization Neural Networks

    T. Sasakawa, J. Hu, K. Isono, K. Hirasawa

    IJCNN 2006   Vancouver   9535 - 9540  2006

  • Support Vector Machine with Fuzzy Decision-Making for Real-world Data Classification

    B. Li, J. Hu, K. Hirasawa, P. Sun, K. Marko

    IJCNN 2006   Vancouver   1314 - 1319  2006

  • Service Area-based Elevator Group Supervisory Control System Using GNP with RL

    J. Zhou, L. Yu, S. Mabu, K. Hirasawa, J. Hu, S. Markon

    SICE 2006   Pusan   5067 - 5072  2006

  • A Double-deck Elevator Group Supervisory Control System with Destination Floor Guidance System using Genetic Network Programming

    L. Yu, J. Zhou, S. Mabu, K. Hirasawa, J. Hu

    SICE 2006   Pusan   5089 - 5094  2006

  • Genetic Network Programming Considering the Evolution of Breadth and Depth

    S. Eto, S. Mabu, K. Hirasawa, J. Hu

    SICE 2006   Pusan   5504 - 5508  2006

  • Constrained Global Optimization Problem by RasID-GA

    D. Sohn, S. Mabu, K. Hirasawa, J. Hu

    SICE 2006   Pusan   5509 - 5514  2006

  • Elevator Group Supervisory Control System with Destination Floor Guidance System using Genetic Network Programming

    H. Heo, S. Mabu, K. Hirasawa, J. Hu

    SICE 2006   Pusan   5489 - 5493  2006

    DOI

  • Self-Adaptive Mechanism in Genetic Network Programming for Mining Association Rules

    K. Taboada, K. shimada, S. Mabu, K. Hirasawa, J. Hu

    SICE 2006   Pusan   6007 - 6012  2006

  • Fuzzy decision-making SVM with an offset for real-world lopsided data classification

    Boyang Li, Jinglu Hu, Kotaro Hirasawa

    2006 SICE-ICASE International Joint Conference   Pusan   143 - 148  2006

     View Summary

    An improved support vector machine (SVM) classifier model for classifying the real-world lopsided data is proposed. The most obvious differences between the model proposed and conventional SVM classifiers are the designs of decision-making functions and the introduction of an offset parameter. With considering about the vagueness of the real-world data sets, a fuzzy decision-making function is designed to take the place of the traditional sign function in the prediction part of SVM classifier. Because of the existence of the interaction and noises influence around the boundary between different clusters, this flexible design of decision-making model which is more similar to the real-world situations can present better performances. In addition, in this paper we mainly discuss an offset parameter introduced to modify the boundary excursion caused by the imbalance of the real-world datasets. Because noises in the real-world can also influence the separation boundary, a weighted harmonic mean (WHM) method is used to modify the offset parameter. Due to these improvements, more robust performances are presented in our simulations. © 2006 ICASE.

    DOI

  • Class Association Rule Mining with Chi-Squared Test Using Genetic Network Programming

    K. Shimada, K. Hirasawa, J. Hu

    IEEE SMC 2006   Taipei   5338 - 5344  2006

    DOI

  • Elevator Group Supervisory Control System Using Genetic Network Programming Considering Ranking Calculation and Node Function Optimization

    EGUCHI Toru, ZHOU Jin, HIRASAWA Kotaro, FURUTSUKI Takayuki, MARKON Sandor

      42 ( 3 ) 281 - 290  2006

    DOI CiNii

  • Genetic Network Programming with Automatically Generated Macro Nodes of Variable Size

    MABU Shingo, HATAKEYAMA Hiroyuki, NAKAGOE Hiroshi, HIRASAWA Kotaro, FURUZUKI Takayuki

    IEEJ Transactions on Electronics, Information and Systems   126 ( 4 ) 548 - 555  2006

     View Summary

    Recently, Genetic Network Programming (GNP) has been proposed as one of the evolutionary algorithms. It represents its solutions as directed graph structures and the distinguished abilities have been shown. However, when we apply GNP to complex problems like the real world one, GNP must have robustness against the changes of environments and evolve quickly. Therefore, we introduced Automatically Generated Macro Nodes (AGMs) to GNP (GNP with AGMs). Actually GNP with AGMs has shown higher performances than the conventional GNP in terms of the fitness and the speed of evolution. In this paper, a new mechanism, AGMs with variable size, is introduced to GNP. Conventional AGMs have the fixed number of nodes and they evolve using only genetic operations, while a new method allows AGM to add nodes by necessity and delete nodes which do not contribute to the evolution of the AGM. The proposed GNP with AGMs of variable size is expected to evolve effectively and efficiently when it is applied to agent systems and also expected to make better behavior sequences of agents more easily than the conventional GNP algorithm. In the simulations, the proposed and conventional methods are applied to a tileworld problem and they are compared with each other. From the results, GNP with AGMs of variable size shows better fitness than GNP with AGMs of fixed size and the conventional GNP when adapting ten different environments.

    DOI CiNii

  • Trading Rules on the Stock Markets Using Genetic Network Programming with Importance Index

    IZUMI Yoshihiro, HIRASAWA Kotaro, HU Jinglu

      42 ( 5 ) 559 - 566  2006

    DOI CiNii

  • 遺伝的ネットワークプログラミングを用いた医療相関ルールの抽出

    嶋田香, 王若慎, 平澤宏太郎, 古月敬之

    電気学会論文誌 C   126 ( 7 ) 849 - 856  2006

    DOI CiNii

  • Genetic network programming with reinforcement learning and its application to making mobile robot behavior

    Shingo Mabu, Hiroyuki Hatakeyamay, Moe Thu Thu, Kotaro Hirasawa, Jinglu Hu

    IEEJ Transactions on Electronics, Information and Systems   126 ( 8 ) 1009 - 1015  2006

     View Summary

    A new graph-based evolutionary algorithm called "Genetic Network Programming, GNP" has been proposed. The solutions of GNP are represented as graph structures, which can improve the expression ability and performance. In addition, GNP with Reinforcement Learning (GNP-RL) has been proposed to search for solutions efficiently. GNP-RL can use current information (state and reward) and change its programs during task execution. Thus, it has an advantage over evolution-based algorithms in case much information can be obtained during task execution. The GNP we proposed in the previous research deals with discrete information, but in this paper, we extend the conventional GNP-RL which can deal with numerical information. The proposed method is applied to the controller of Khepera simulator and its performance is evaluated.

    DOI CiNii

  • A Brain-like Learning System with Supervised, Unsupervised and Reinforcement Learning

    SASAKAWA Takafumi, HU Jinglu, HIRASAWA Kotaro

    IEEJ Transactions on Electronics, Information and Systems   126 ( 9 ) 1165 - 1172  2006

     View Summary

    Our brain has three different learning paradigms: supervised, unsupervised and reinforcement learning. And it is suggested that those learning paradigms relate deeply to the cerebellum, cerebral cortex and basal ganglia in the brain, respectively. Inspired by these knowledge of brain, we present a brain-like learning system with those three different learning algorithms. The proposed system consists of three parts: the supervised learning (SL) part, the unsupervised learning (UL) part and the reinforcement learning (RL) part. The SL part, corresponding to the cerebellum of brain, learns an input-output mapping by supervised learning. The UL part, corresponding to the cerebral cortex of brain, is a competitive learning network, and divides an input space to subspaces by unsupervised learning. The RL part, corresponding to the basal ganglia of brain, optimizes the model performance by reinforcement learning. Numerical simulations show that the proposed brain-like learning system optimizes its performance automatically and has superior performance to an ordinary neural network.

    DOI CiNii

  • 重要度指標付きGenetic Network Programmingにおける機能切り替えについて

    江藤慎治, 畠山裕之, 間普真吾, 平澤宏太郎, 古月敬之

    情報処理学会論文誌   47 ( 9 ) 2860 - 2868  2006

  • Optimization of Double-deck Elevator Group Supervisory Control System using Genetic Network Programming

    EGUCHI Toru, ZHOU Jin, HIRASAWA Kotaro, FURUTSUKI Takayuki, MARKON Sandor

    Transactions of the Society of Instrument and Control Engineers   42 ( 11 ) 1260 - 1268  2006

     View Summary

    In recent years, double-deck elevator systems (DDES) where two cars are connected in a shaft has been developed for the rising demand of more efficient transport of passengers in high-rise buildings. DDES has specific behaviors due to the connection of cars and the need for securing comfortable riding, so its group control becomes more complex than conventional single-deck elevator systems (SDES). Meanwhile, a graph-based evolutionary method, Genetic Network Programming (GNP) has been applied to elevator group supervisory control systems, and its effectiveness is clarified. GNP can consider the specific behaviors of DDES in its node functions easily and execute an efficient rule-based group control optimized evolutionary. In this paper, a new group control system for DDES using GNP is proposed, and its optimization and performance verification are done through the simulations. First, optimization of GNP for DDES is executed. Second, the performance of the proposed method is verified by the comparison with conventional methods, and the obtained control rules are studied. Finally, the performance improvement by the proposal is evaluated in terms of SDES capacity.

    DOI CiNii

  • Association Rule Mining Using Genetic Network Programming

    SHIMADA Kaoru, HIRASAWA Kotaro, FURUZUKI Takayuki

    J. SOFT   18 ( 6 ) 881 - 891  2006

     View Summary

    A method of association rule mining is proposed using Genetic Network Programming (GNP) to improve the performance of rule extraction. Association rule mining is the discovery of association relationships or correlations among a set of attributes in a database. GNP is one of the evolutionary optimization techniques, which uses directed graph structures as genes. GNP examines the attribute values of database tuples using judgement nodes and calculates the measurements of association rules using processing nodes. In addition, the proposed method measures the significance of associations via the chi-squared test for correlation used in classical statistics using GNP's feature. Extracted association rules are stored in a pool all together through the generations in order to find new important rules. Therefore, the proposed method is fundamentally different from the previous methods in its evolutionary way. In this paper, the algorithm capable of finding the important association rules is described and some experimental results are shown.

    DOI CiNii

  • Genetic Network Programming with Acquisition Mechanism of Association Rules

    K.Shimada, K.Hirasawa, J.Hu

    Journal of Advanced Computational Intelligence and Intelligent Informatics   10 ( 1 ) 102 - 111  2006

  • Elevator Group Supervisory Control System Using Genetic Network Programming with Functional Localization

    T. Eguchi, J. Zhou, S. Eto, K. Hirasawa, J. Hu, S. Markon

    Jounal of Advanced Computational Intelligence and Intelligent Informatics   10 ( 3 ) 385 - 393  2006

  • Realizing Functional Localization Using Genetic Network Programming with Important Index

    S. Eto, H. Hatakeyama, S. Mabu, K. Hirasawa, J. Hu

    Journal of Advanced Computaional Intelligence and Intelligent Informatics   10 ( 4 ) 555 - 566  2006

  • Alternate Genetic Network Programming with Associate Rule Acquisition Mechanism Between Attribute Families

    K. Shimada, K. Hirasawa, J. Hu

    Journal of Advanced computational Intelligence and Intelligent Informatics   10 ( 6 ) 954 - 963  2006

  • Adaptive Random Search with Intesification and Diversification Combined with Genetic Algorithm

    D. Sohn, H. Hatakeyama, S. Mabu, K. Hirasawa, J. Hu

    Journal of Advanced Computational Intelligence and Intelligent Informatics   10 ( 6 ) 921 - 930  2006

  • Benchmark Test of RasID-GA for Inequality/Equality Constrained Optimization

    D. Sohn, S. Mabu, K. Hirasawa, J. Hu

    インテリジェントシステムシンポジウム   柏   155 - 160  2006

  • An Improved Method for Identification of Quasi-ARMAX Model

    J. Ji, J. Hu, K. Hirasawa

    SICE九州支部学術講演会   佐賀   43 - 44  2006

  • Evolutionary Method of Genetic Network Programming Considering Breadth and Depth

    S. Eto, S. Mabu, K. Hirasawa

    GECCO 2006   Seattle  2006

  • Associate Rule Mining with Chi-Squared Test Using Alternate Genetic Network Programming

    K. Shimada, K. Hirasawa, J. Hu

    ICDM 2006   Leipzig   202 - 216  2006

    DOI

  • Genetic Network Programming with Reinforcement Learning Using Sarsa Algorithm

    S. Mabu, H. Hatakeyama, K. Hirasawa, J. Hu

    CEC 2006   Vancouver   1570 - 1575  2006

  • An Extension of Genetic Network Programming with Reinforcement Learning Using Actor-Critic

    H. Hatakeyama, S. Mabu, K. Hirasawa, J. Hu

    CEC 2006   Vancouver   5686 - 5692  2006

  • Trading Rules on the Stock Markets using Genetic Network Programming with Candlestick Chart

    Y. Izumi, T. Yamaguchi, S. Mabu, K. Hirasawa, J. Hu

    CEC 2006   Vancouver   8531 - 8536  2006

  • RasID-GA with Simplex Crossover for Optimization Problems

    D. Sohn, S. Mabu, K. Hirasawa, J. Hu

    CEC 2006   Vancouver   10378 - 10385  2006

  • A study of applying Genetic Network Programming with Reinforcement Learning to Elevator Group Supervisory Control System

    J. Zhou, T. Eguchi, S. Mabu, K. Hirasawa, J. Hu, S. Markon

    CEC 2006   Vancouver   10392 - 10398  2006

  • Effective Training Method for Functional Localization Neural Networks

    T. Sasakawa, J. Hu, K. Isono, K. Hirasawa

    IJCNN 2006   Vancouver   9535 - 9540  2006

  • Support Vector Machine with Fuzzy Decision-Making for Real-world Data Classification

    B. Li, J. Hu, K. Hirasawa, P. Sun, K. Marko

    IJCNN 2006   Vancouver   1314 - 1319  2006

  • Service Area-based Elevator Group Supervisory Control System Using GNP with RL

    J. Zhou, L. Yu, S. Mabu, K. Hirasawa, J. Hu, S. Markon

    SICE 2006   Pusan   5067 - 5072  2006

  • A Double-deck Elevator Group Supervisory Control System with Destination Floor Guidance System using Genetic Network Programming

    L. Yu, J. Zhou, S. Mabu, K. Hirasawa, J. Hu

    SICE 2006   Pusan   5089 - 5094  2006

  • Genetic Network Programming Considering the Evolution of Breadth and Depth

    S. Eto, S. Mabu, K. Hirasawa, J. Hu

    SICE 2006   Pusan   5504 - 5508  2006

  • Constrained Global Optimization Problem by RasID-GA

    D. Sohn, S. Mabu, K. Hirasawa, J. Hu

    SICE 2006   Pusan   5509 - 5514  2006

  • Self-Adaptive Mechanism in Genetic Network Programming for Mining Association Rules

    K. Taboada, K. shimada, S. Mabu, K. Hirasawa, J. Hu

    SICE 2006   Pusan   6007 - 6012  2006

  • Fuzzy Decision-making SVM with An Offset for Realーworld Lopsided Data Classification

    B. Li, J. Hu, K. Hirasawa

    SICE 2006   Pusan   143 - 148  2006

  • Class Association Rule Mining with Chi-Squared Test Using Genetic Network Programming

    K. Shimada, K. Hirasawa, J. Hu

    IEEE SMC 2006   Taipei   5338 - 5344  2006

    DOI

  • Genetic network programming with reinforcement learning and its application to making mobile robot behavior

    Shingo Mabu, Hiroyuki Hatakeyamay, Moe Thu Thu, Kotaro Hirasawa, Jinglu Hu

    IEEJ Transactions on Electronics, Information and Systems   126 ( 8 ) 1009 - 1015  2006

     View Summary

    A new graph-based evolutionary algorithm called "Genetic Network Programming, GNP" has been proposed. The solutions of GNP are represented as graph structures, which can improve the expression ability and performance. In addition, GNP with Reinforcement Learning (GNP-RL) has been proposed to search for solutions efficiently. GNP-RL can use current information (state and reward) and change its programs during task execution. Thus, it has an advantage over evolution-based algorithms in case much information can be obtained during task execution. The GNP we proposed in the previous research deals with discrete information, but in this paper, we extend the conventional GNP-RL which can deal with numerical information. The proposed method is applied to the controller of Khepera simulator and its performance is evaluated.

    DOI CiNii

  • Genetic Network Programming with Acquisition Mechanism of Association Rules

    K.Shimada, K.Hirasawa, J.Hu

    Journal of Advanced Computational Intelligence and Intelligent Informatics   10 ( 1 ) 102 - 111  2006

  • Elevator Group Supervisory Control System Using Genetic Network Programming with Functional Localization

    T. Eguchi, J. Zhou, S. Eto, K. Hirasawa, J. Hu, S. Markon

    Jounal of Advanced Computational Intelligence and Intelligent Informatics   10 ( 3 ) 385 - 393  2006

  • Realizing Functional Localization Using Genetic Network Programming with Important Index

    S. Eto, H. Hatakeyama, S. Mabu, K. Hirasawa, J. Hu

    Journal of Advanced Computaional Intelligence and Intelligent Informatics   10 ( 4 ) 555 - 566  2006

  • Alternate Genetic Network Programming with Associate Rule Acquisition Mechanism Between Attribute Families

    K. Shimada, K. Hirasawa, J. Hu

    Journal of Advanced computational Intelligence and Intelligent Informatics   10 ( 6 ) 954 - 963  2006

  • Adaptive Random Search with Intesification and Diversification Combined with Genetic Algorithm

    D. Sohn, H. Hatakeyama, S. Mabu, K. Hirasawa, J. Hu

    Journal of Advanced Computational Intelligence and Intelligent Informatics   10 ( 6 ) 921 - 930  2006

  • 機能局在型Genetic Network Programmingによる機能切り替えの進化

    江藤慎治, 平澤宏太郎, 古月敬之

    情報処理学会 火の国情報シンポジウム   飯塚  2005

  • 獲得情報を用いる遺伝的ネットワークプログラミングによるデータマイニング

    嶋田薫, 王若しん, 平澤宏太郎, 古月敬之

    情報処理学会 火の国情報シンポジウム   飯塚  2005

  • Genetic Network Programmingを用いた株式売買ポートフォリオモデルの構築

    泉良祐, 高橋好史, 平澤宏太郎, 古月敬之

    人口頭脳シンポジウム   佐賀   42 - 43  2005

  • 学習と進化を用いた株式売買モデルの構築

    高橋好史, 泉良裕, 平澤宏太郎, 古月敬之

    人口頭脳シンポジウム   佐賀   56 - 57  2005

  • マルチブランチニューラルネットワークによる株価予測

    山下貴志, 平澤宏太郎, 古月敬之

    電気学会C部門大会   北九州   1066 - 1071  2005

  • 遺伝的ネットワークプログラミングを用いた医療相関ルールの抽出

    嶋田香, 王若しん, 平澤宏太郎, 古月敬之

    電気学会C部門大会   北九州   995 - 1000  2005

  • A Study of Genetic Network Programming with Reinforcement Learning and its Application

    S. Mabu, M. Thu Thu, K. Hirasawa, J. Hu

    電気学会C部門大会   北九州   1019 - 1024  2005

  • 機能局在型ニューラルネットワークの学習の一方法

    磯野功典, 笹川隆史, 古月敬之, 平澤宏太郎

    電気学会C部門大会   北九州   1056 - 1057  2005

  • 交替型遺伝的ネットワークプログラミングを用いた2つの属性グループからの相関ルールの抽出

    嶋田香, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   京都   341 - 346  2005

  • 重要度指標付Genetic Network Programmingを用いた株式売買モデル

    泉良祐, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   京都   335 - 340  2005

  • 重要度指標付機能局在型Genetic Network Programmingの機能切り替えについて

    江藤慎治, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   京都   73 - 78  2005

  • Actor-Criticを用いた遺伝的ネットワークプログラミングの学習アルゴリズム

    間普真吾, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   京都   381 - 386  2005

  • Adaptive Random Search with Intensification and Diversification combined with Genetic Algorithm

    D. Sohn, K. Hirasawa, J. Hu

    インテリジェントシステムシンポジウム   京都   67 - 72  2005

  • Genetic Network Programming with Reinforcement Learning を用いた株式売買モデル

    高橋好史, 平澤宏太郎, 古月敬之

    インテリジェントシステムシンポジウム   京都   387 - 390  2005

  • Improving the Tuning Capability of the Adjusting Neural Network

    Y. Sugita, K. Hirasawa

    AROB 2004   Beppu  2005

  • Association Rule Mining Using Genetic Network Programming

    K. Shimada, K. Hirasawa, T. Furutsuki

    AROB 2004   Beppu  2005

  • Data Mining using Genetic Network Programming

    T. Fukuda, K. Shimada, K. Hirasawa, T. Furutsuki

    AROB2004   Beppu  2005

  • Performance optimization of function localization neural network by using reinforcement learning

    Takafumi Sasakawa, Jinglu Hu, Kotaro Hirasawa

    Proceedings of the International Joint Conference on Neural Networks   2   1314 - 1319  2005

     View Summary

    According to Hebb's cell assembly theory, the brain has the capability of function localization. On the other hand, it is suggested that the brain has three different learning paradigms: supervised, unsupervised and reinforcement learning. Inspired by the above knowledge of brain, we present a selforganizing function localization neural network (FLNN), that contains supervised, unsupervised and reinforcement learning paradigms. In this paper, we concentrate our discussion mainly on applying a simplified reinforcement learning called evaluative feedback to optimization of the self-organizing FLNN. Numerical simulations show that the self-organizing FLNN has superior performance to an ordinary artificial neural network (ANN). © 2005 IEEE.

    DOI

  • Application of Multi-Branch Neural Networks to Stock Market Prediction

    T. Yamasita, K. Hirasawa, J. Hu

    IJCNN 2005   Montreal   2544 - 2548  2005

    DOI

  • A Neural Network Approach to Improving Identification of Nonlinear Polynominal Models

    J. Hu, Y. Li, K. Hirasawa

    SICE 2005   Okayama   1662 - 1667  2005

  • Genetic Network Programming with Actor-Critic and its Application

    S. Mabu, K. Hirasawa, J. Hu

    SICE 2005   Okayama   3635 - 3640  2005

  • Adaptive Random Search with Intensification and Diversification combined with Genetic Algorithm using Automatic Switching

    D. Sohn, K. Hirasawa, J. Hu

    SICE 2005   Okayama   28 - 34  2005

  • Symbiotic Learning and Evolution on Social Science

    N. Ota, K. Hirasawa, J. Hu

    SICE 2005   Okayama   3649 - 3652  2005

  • Elevator Group Supervisory Control using GNP with Reinforcement Learning based on Normalized Information

    J. Zhou, K. Hirasawa, J. Hu, S. Markon

    SICE 2005   Okayama   74 - 79  2005

  • Elevator Group Supervisory Control System Using Genetic Network Programming

    T. Eguchi, K. Hirasawa, J. Hu, S. Markon

    SICE 2005   Okayama   1 - 6  2005

  • Genetic Network Programming with Association Rule Acquisition Mechanisms

    K. Shimada, K. Hirasawa, J. Hu

    SICE 2005   Okayama   13 - 18  2005

  • Genetic Network Programming with Functional Localization

    S. Eto, K. Hirasawa, J. Hu

    SICE 2005   Okayama   20 - 27  2005

  • Time Series Prediction System of Stock Price using Multi-Branch Neural Networks

    T. Yamasita, K. Hirasawa, J. Hu

    SICE 2005   Okayama   2057 - 2062  2005

  • Elevator Group Supervisory Control System Using Genetic Network Programming with Functional Localization

    T. Eguchi, K. Hirasawa, J. Hu

    CEC 2005   Edinbburgh   328 - 335  2005

  • Adaptive Random Search with Intensification and Diversification combined with Genetic Algorithm

    D. Sohn, K. Hirasawa, J. Hu

    CEC 2005   Edinburgh   1462 - 1467  2005

  • Elevator Group Supervisory Control System Using Genetic Network Programming with Reinforccment Learning

    J. Zhou, T. Eguchi, K. Hirasawa, J. Hu

    CEC 2005   Edinburgh   336 - 342  2005

  • Multi-Branch Neural Networks and its Application to Stock Price Prediction

    T. Yamasita, K. Hirasawa, J. Hu

    KES 2005   Melbourne   1 - 7  2005

  • Genetic Network Programming with Acquisition Mechanisms of Association Rules in Dense Database

    K. Shimada, K. Hirasawa, J. Hu

    CIMCA 2005   Vienna  2005

  • Switching for Functional Localization of Genetic Network Programming

    S. Eto, K. Hirasawa, J. Hu

    ICMLA 2005   Los Angeles   325 - 330  2005

    DOI

  • Self-organizing Function Localization Neural Network

    SASAKAWA Takafumi, HU Jinglu, HIRASAWA Kotaro

    Transactions of the Society of Instrument and Control Engineers   41 ( 1 ) 67 - 74  2005

     View Summary

    In an ordinary artificial neural network, individual neurons do not have any special relations with input patterns. That is, an ordinary neural network has only learning capability, but does not have the capability of function localization. However, according to Hebb's cell assembly theory about how the brain worked, it is suggested that it has function localization in the brain, which means that specific groups of neurons are activated corresponding to certain sorts of sensory information the brain receives. On the other hand, it is also reported that the cerebellum and cerebral cortex in the brain are specialized in supervised and unsupervised learning paradigms, respectively. Inspired by both Hebb's cell assembly theory, and the basic learning paradigms in the brain, this paper presents a self-organizing function localization neural network (FLNN). The proposed self-organizing FLNN consists of two parts: the main part and the control part. The main part, corresponding to the cerebellum of brain, is an ordinary 3-layered feedforward neural network, but each hidden neuron contains a signal from the control part, controlling its firing strength. The control part, corresponding to the cerebral cortex of brain, consists of a self-organizing map (SOM) network whose outputs are associated with the hidden neurons of the main part. Trained with an unsupervised learning algorithm, the SOM control part extracts structural features of input space and controls the firing strength of hidden neurons in the main part. And the main part realizes an input-output mapping by using supervised learning. In this way, the self-organizing FLNN realizes the capabilities of both function localization and learning. Numerical simulations show that the self organizing FLNN has superior performance to an ordinary neural network.

    DOI CiNii

  • 機能局在型Genetic Network Programmingの構成

    江藤慎治, 平澤宏太郎, 古月敬之

    電気学会論文誌 C   125 ( 2 ) 329 - 336  2005

    DOI CiNii

  • Genetic Network Programmingによる株価予測と売買モデル

    森茂男, 平澤宏太郎, 古月敬之

    電気学会論文誌C   125 ( 4 ) 631 - 636  2005

    DOI CiNii

  • 階層型ニューラルネットワークにおけるマルチブランチ構造とその局所性

    山下貴志, 平澤宏太郎, 古月敬之

    電気学会論文誌C   125 ( 6 ) 941 - 947  2005

    DOI CiNii

  • Genetic Network Programmingによるエレベータ群管理システムの基礎検討

    江口徹, 周金, 平澤宏太郎, 古月敬之, マルコン シャンドル

    電気学会論文誌C   125 ( 7 ) 1055 - 1062  2005

    DOI CiNii

  • Elevator Group Control Using Multiagent Task-Oriented Reinforcement Learning

    M. A.S. Kamal, Junichi Murata, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   125 ( 7 ) 1140 - 1146  2005

     View Summary

    In this paper, a reinforcement learning method is proposed that optimizes passenger service in elevator group systems. Task-oriented reinforcement learning using multiple agents is applied in the control system in allocating immediate landing calls to the elevators and operating them intelligently in attaining better service in this stochastic dynamic domain. The proposed system shows better adaptive performance in different traffic profiles with faster convergence compared to the other learning elevator group control system. © 2005, The Institute of Electrical Engineers of Japan. All rights reserved.

    DOI CiNii

  • 獲得した情報を用いる遺伝的ネットワークプログラミングによるデータマイニング

    嶋田香, 平澤宏太郎, 古月敬之

    情報処理学会論文誌   46 ( 10 ) 2576 - 2586  2005

  • Multi-Branch Neural Networks with Functional Localization by Branch Control

    YAMASHITA Takashi, HIRASAWA Kotaro, FURUZUKI Takayuki

    Journal of Japan Society for Fuzzy Theory and Intelligent Informatics   17 ( 5 ) 108 - 115  2005

     View Summary

    Neural networks (NNs) can solve only a simple problem if the network size is too small. On the other hand, if the network size increases, it costs a lot in terms of memory space andcalculation time. Therefore, we have studied how to construct the network structure with high performances and low costs in space and time. A solution is a multi-branch structure. Conventional NNs use the single-branch for the connections, while the multi-branch structurehas multibranches between nodes. In this paper, a new method which enables the multi-branch NNs to have functional localization is proposed. Neural networks with Branch Control adjust signals propagating through branches between the intermediate layer and output layer depending on the inputs of the network. Therefore, a branch could be cut depending on input values. Simulation results of function approximations and a classification problem illustrated the effectiveness of multi-branch NNs with functional localization.

    DOI CiNii

  • 強化学習を用いた遺伝的ネットワークプログラミングとそのエージェントの行動生成における性能評価

    間普真吾, 平澤宏太郎, 古月敬之

    情報処理学会論文誌   46 ( 12 ) 3207 - 3217  2005

  • Genetic Network Programming for Automatic Program Generation

    S. Mabu, K. Hirasawa, T. Matsuya, J. Hu

    J. of Advanced Computational Intelligence and Intelligent Informatics   9 ( 4 ) 430 - 436  2005

  • A Study of Genetic Network Programming with Reinforcement Learning and its Application

    S. Mabu, M. Thu Thu, K. Hirasawa, J. Hu

    電気学会C部門大会   北九州   1019 - 1024  2005

  • Adaptive Random Search with Intensification and Diversification combined with Genetic Algorithm

    D. Sohn, K. Hirasawa, J. Hu

    インテリジェントシステムシンポジウム   京都   67 - 72  2005

  • Improving the Tuning Capability of the Adjusting Neural Network

    Y. Sugita, K. Hirasawa

    AROB 2004   Beppu  2005

  • Association Rule Mining Using Genetic Network Programming

    K. Shimada, K. Hirasawa, T. Furutsuki

    AROB 2004   Beppu  2005

  • Data Mining using Genetic Network Programming

    T. Fukuda, K. Shimada, K. Hirasawa, T. Furutsuki

    AROB2004   Beppu  2005

  • Performance optimization of function localization neural network by using reinforcement learning

    Takafumi Sasakawa, Jinglu Hu, Kotaro Hirasawa

    Proceedings of the International Joint Conference on Neural Networks   2   1314 - 1319  2005

     View Summary

    According to Hebb's cell assembly theory, the brain has the capability of function localization. On the other hand, it is suggested that the brain has three different learning paradigms: supervised, unsupervised and reinforcement learning. Inspired by the above knowledge of brain, we present a selforganizing function localization neural network (FLNN), that contains supervised, unsupervised and reinforcement learning paradigms. In this paper, we concentrate our discussion mainly on applying a simplified reinforcement learning called evaluative feedback to optimization of the self-organizing FLNN. Numerical simulations show that the self-organizing FLNN has superior performance to an ordinary artificial neural network (ANN). © 2005 IEEE.

    DOI

  • Application of Multi-Branch Neural Networks to Stock Market Prediction

    T. Yamasita, K. Hirasawa, J. Hu

    IJCNN 2005   Montreal   2544 - 2548  2005

    DOI

  • A Neural Network Approach to Improving Identification of Nonlinear Polynominal Models

    J. Hu, Y. Li, K. Hirasawa

    SICE 2005   Okayama   1662 - 1667  2005

  • Genetic Network Programming with Actor-Critic and its Application

    S. Mabu, K. Hirasawa, J. Hu

    SICE 2005   Okayama   3635 - 3640  2005

  • Adaptive Random Search with Intensification and Diversification combined with Genetic Algorithm using Automatic Switching

    D. Sohn, K. Hirasawa, J. Hu

    SICE 2005   Okayama   28 - 34  2005

  • Symbiotic Learning and Evolution on Social Science

    N. Ota, K. Hirasawa, J. Hu

    SICE 2005   Okayama   3649 - 3652  2005

  • Elevator Group Supervisory Control using GNP with Reinforcement Learning based on Normalized Information

    J. Zhou, K. Hirasawa, J. Hu, S. Markon

    SICE 2005   Okayama   74 - 79  2005

  • Elevator Group Supervisory Control System Using Genetic Network Programming

    T. Eguchi, K. Hirasawa, J. Hu, S. Markon

    SICE 2005   Okayama   1 - 6  2005

  • Genetic Network Programming with Association Rule Acquisition Mechanisms

    K. Shimada, K. Hirasawa, J. Hu

    SICE 2005   Okayama   13 - 18  2005

  • Genetic Network Programming with Functional Localization

    S. Eto, K. Hirasawa, J. Hu

    SICE 2005   Okayama   20 - 27  2005

  • Time Series Prediction System of Stock Price using Multi-Branch Neural Networks

    T. Yamasita, K. Hirasawa, J. Hu

    SICE 2005   Okayama   2057 - 2062  2005

  • Elevator Group Supervisory Control System Using Genetic Network Programming with Functional Localization

    T. Eguchi, K. Hirasawa, J. Hu

    CEC 2005   Edinbburgh   328 - 335  2005

  • Adaptive Random Search with Intensification and Diversification combined with Genetic Algorithm

    D. Sohn, K. Hirasawa, J. Hu

    CEC 2005   Edinburgh   1462 - 1467  2005

  • Elevator Group Supervisory Control System Using Genetic Network Programming with Reinforccment Learning

    J. Zhou, T. Eguchi, K. Hirasawa, J. Hu

    CEC 2005   Edinburgh   336 - 342  2005

  • Multi-Branch Neural Networks and its Application to Stock Price Prediction

    T. Yamasita, K. Hirasawa, J. Hu

    KES 2005   Melbourne   1 - 7  2005

  • Genetic Network Programming with Acquisition Mechanisms of Association Rules in Dense Database

    K. Shimada, K. Hirasawa, J. Hu

    CIMCA 2005   Vienna  2005

  • Switching for functional localization of genetic network programming

    Shinji Eto, Kotaro Hirasawa, Jingle Hu

    Proceedings - ICMLA 2005: Fourth International Conference on Machine Learning and Applications   2005   325 - 330  2005

     View Summary

    Many methods of generating behavior sequences of agents by evolution have been reported. A new evolutionary computation method named Genetic Network Programming (GNP) has also been developed recently along with these trends. The aim of this paper is to build an artificial model to realize functional localization based on GNP considering the fact that the functional localization of the brain is realized in such a way that a different part of the brain corresponds to a different function. GNP has a directed graph structure suitable for realizing functional localization. In this paper, it is especially stated that the evolution of the switching function can be realized for functional localization of GNP using the self-sufficient garbage collector problem. © 2005 IEEE.

    DOI

  • Elevator Group Control Using Multiagent Task-Oriented Reinforcement Learning

    M. A.S. Kamal, Junichi Murata, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   125 ( 7 ) 1140 - 1146  2005

     View Summary

    In this paper, a reinforcement learning method is proposed that optimizes passenger service in elevator group systems. Task-oriented reinforcement learning using multiple agents is applied in the control system in allocating immediate landing calls to the elevators and operating them intelligently in attaining better service in this stochastic dynamic domain. The proposed system shows better adaptive performance in different traffic profiles with faster convergence compared to the other learning elevator group control system. © 2005, The Institute of Electrical Engineers of Japan. All rights reserved.

    DOI CiNii

  • Genetic Network Programming for Automatic Program Generation

    S. Mabu, K. Hirasawa, T. Matsuya, J. Hu

    J. of Advanced Computational Intelligence and Intelligent Informatics   9 ( 4 ) 430 - 436  2005

  • An adaptive state estimator for pulverizer control using moments of particle size distribution

    Y Fukayama, K Hirasawa, K Shimohira, H Kanemoto

    IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY   12 ( 6 ) 797 - 810  2004.11

     View Summary

    An adaptive state estimator for pulverizers consisting of blending, grinding, and classifying processes has been developed in order to improve control of pulverized-coal-fired power stations. Though coal flow and non-Gaussian particle size distributions in the processes are mutually related, the estimator is able to efficiently simulate flow and normalized moments of the distributions with a state vector. The estimator also identifies coal grindability for adapting to variation in coal characteristic in parallel with the process simulation. The accuracy of the adaptive estimation and the effectiveness in improving the load-swinging performance have been validated at a 1000-MWe class power station.

    DOI CiNii

  • An adaptive state estimator for pulverizer control using moments of particle size distribution

    Y Fukayama, K Hirasawa, K Shimohira, H Kanemoto

    IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY   12 ( 6 ) 797 - 810  2004.11

     View Summary

    An adaptive state estimator for pulverizers consisting of blending, grinding, and classifying processes has been developed in order to improve control of pulverized-coal-fired power stations. Though coal flow and non-Gaussian particle size distributions in the processes are mutually related, the estimator is able to efficiently simulate flow and normalized moments of the distributions with a state vector. The estimator also identifies coal grindability for adapting to variation in coal characteristic in parallel with the process simulation. The accuracy of the adaptive estimation and the effectiveness in improving the load-swinging performance have been validated at a 1000-MWe class power station.

    DOI CiNii

  • 強化学習を用いた遺伝的ネットワークプログラミングの学習法とその性能評価

    間普真吾, 平澤宏太郎, 古月敬之

    情報処理学会 火の国情報シンポジウム   大分  2004

  • プログラムサイズ可変型マクロノード付遺伝的ネットワークプログラミング

    中越洋, 平澤宏太郎, 古月隆之

    ファジイシンポジウム   北九州   607 - 610  2004

  • マルチゲート付ニューラルネットワーク

    後藤健一, 平澤宏太郎, 古月隆之

    ファジイシンポジウム   北九州   13 - 18  2004

  • 階層型ニューラルネットワークにおけるマルチブランチ構造とその局所性

    山下貴志, 平澤宏太郎, 古月敬之

    ファジイシンポジウム   北九州   13 - 18  2004

  • 自己組織化機能局在型ニューラルネットワーク

    笹川隆史, 古月隆之, 平澤宏太郎

    ファジイシンポジウム   北九州   27 - 30  2004

  • 将棋棋譜自動生成システムのための駒認識方式

    妻鹿大祐, 浜田長晴

    画像電子学会年次大会   東京  2004

    DOI

  • Genetic Network Programming によるエレベータ群管理システムの評価

    江口徹, 平澤宏太郎, 古月敬之, マルコンシャンドル

    電気学会C部門大会   宇都宮   526 - 531  2004

  • 遺伝的ネットワークプログラミングによる相関ルールの抽出

    嶋田薫, 平澤宏太郎, 古月敬之

    FANシンポジウム   高知   363 - 368  2004

  • 進化と学習を用いた遺伝的ネットワークプログラミングによるエージェントの行動学習

    間普真吾, 平澤宏太郎, 古月敬之

    FANシンポジウム   高知   349 - 354  2004

  • 脳の知見を参考にした機能局在型ニューラルネットワーク

    笹川隆史, 古月敬之, 平澤宏太郎

    FANシンポジウム   高知   91 - 94  2004

  • Genetic Network Programmingによるエレベータ群管理システムの構築

    江口徹, 平澤宏太郎, 古月敬之, マルコンシャンドル

    SICEシステム情報部門学術講演会   浜松   73 - 78  2004

  • 非線形システム解析のためのデータベース構築

    李いんじつ, 古月敬之, 平澤宏太郎

    SICE九州支部学術講演会   北九州   153 - 154  2004

  • 遺伝的適応ランダム探索法(RasID-GA)

    孫東圭, 平澤宏太郎, 古月敬之

    SICE九州支部学術講演会     365 - 368  2004

  • タイルワールド問題を用いたGenetic Network Programmingと従来手法の性能比較

    間普真吾, 平澤宏太郎, 古月敬之

    SICE九州支部学術講演会   北九州   373 - 376  2004

  • ホモトピー連続法を用いた準ARXニューラルネットワークの学習

    ろ, 古月敬之, 平澤宏太郎

    SICE九州支部学術講演会   北九州   141 - 142  2004

  • An Approximate Stability Analysis of a Robust Control DC Motor System

    A. Hussein, K. Hirasawa, J. Hu, K. Wada

    AROB 2003   Oita   333 - 336  2004

  • Evolution of Hybrid Neural Networks Using Genetic Network Programming

    D. Li, K. Hirasawa, J. Hu

    AROB 2003   Oita  2004

  • Genetic algorithm optimization of a convolutional neural network for autonomous crack detection

    R Ouellette, M Browne, K Hirasawa

    CEC2004: PROCEEDINGS OF THE 2004 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1 AND 2   Portland   516 - 521  2004

     View Summary

    Detecting cracks is an important function in building, tunnel, and bridge structural analysis. Successful automation of crack detection can provide a uniform and timely means for preventing further damage to structures. This laboratory has successfully applied convolutional neural networks (CNNs) to online crack detection. CNNs represent an interesting method for adaptive image processing, and form a link between artificial neural networks and finite impulse response filters. As with most artificial neural networks, the CNN is susceptible to multiple local minima, thus complexity and time must be applied in order to avoid becoming trapped within the local minima. This paper employs a standard genetic algorithm (GA) to train the weights of a 4-5x5 filter CNN in order to pass through local minima. This technique resulted in a 92.3 +/- 1.4% average success rate using 25 GA-trained CNNs presented with 100 crack (320x240 pixel) images.

  • Elevator group supervisory control systems using genetic network programming

    T Eguchi, K Hirasawa, J Hu, S Markon

    CEC2004: PROCEEDINGS OF THE 2004 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1 AND 2   Portland   1661 - 1667  2004

     View Summary

    Genetic Network Programming (GNP) has been proposed as a new method of evolutionary computation. Until now, GNP has been applied to various problems and its effectiveness was clarified. However, these problems were virtual models, so the applicability and availability of GNP to the real-world applications have not been studied. In this paper, as a first step of applying GNP to the real-world applications, Elevator Group Supervisory Control Systems (EGSCSs) are considered. Generally, EGSCSs are complex and difficult problems to solve because they are too dynamic and probabilistic. So the design of a useful controller of EGSCSs was very difficult. Recently, the design of such a controller of EGSCSs has been tried actively using Artificial Intelligence (AI) technologies. In this paper, it is reported that the design of a controller of EGSCSs has been studied using GNP whose characteristic is to use directed graph as its gene instead of bit strings and trees of GA and GP From simulations, it is clarified that better solutions are obtained by using GNP than other conventional methods and the availability of GNP to real-world applications is confirmed.

  • Genetic network programming with automatically generated variable size macro nodes

    H Nakagoe, K Hirasawa, J Hu

    CEC2004: PROCEEDINGS OF THE 2004 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1 AND 2   Portland   713 - 719  2004

     View Summary

    Genetic Network Programming (GNP) has directed graph structures as genes, which is extended from other evolutionary computations such as Genetic Algorithm (GA) and Genetic Programming (GP). Generally, macroinstructions are introduced as sub-routines, function localization and so on. Previously, we have introduced the structure of macroinstructions in GNP named Automatically Generated Macro Nodes (AGMs) for reducing the time of evolution efficiently, and showed that macroinstructions are useful to acquire good performances. But the AGMs have fixed number of nodes, and it is found that the effectiveness of evolution of macroinstructions depends on the main program calling them and initialized parameters. Accordingly in this paper, new AGMs are introduced to improve their performances further more by the mechanism of varying the size of AGMs, which are named variable size AGMs. This is the mechanism to add and delete nodes according to necessity. In the simulations, comparisons between GNP program only, GNP with conventional AGMs and GNP with variable size AGMs are carried out using the tile world. Simulation results show that the proposed method is better compared with conventional GNP and GNP with conventional AGMs. And also it is clarified that the node transition rules obtained by new AGMs show the generalized rules able to deal with unknown environments.

  • Functional localization of genetic network programming and its application to a pursuit problem

    SJ Eto, K Hirasawa, J Hu

    CEC2004: PROCEEDINGS OF THE 2004 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1 AND 2   Portland   683 - 690  2004

     View Summary

    According to the knowledge of brain science, it is suggested that there exists cerebral functional localization, which means that a specific part of the cerebrum is activated depending on various kinds of information human receives. The aim of this paper is to build an artificial model to realize functional localization based on Genetic Network Programming (GNP), a new evolutionary computation method recently developed. GNP has a directed graph structure suitable for realizing functional localization. We studied the basic characteristics of the proposed system by making GNP work in a functionally localized way.

  • Genetic Network Programming with Reinforcement Learning and its Performance Evaluation

    S. Mabu, K. Hirasawa, J. Hu

    GECCO 2004   Seattle  2004

  • Self-Organized Functional Localization Neural Network

    T. Sasakawa, J. Hu, K. Hirasawa

    IJCNN 2004   Budapest  2004

    DOI

  • Multi-Branch Structure and Its Localized Property in Layered Neural Networks

    T. Yamasita, K. Hirasawa, J. Hu

    IJCNN 2004   Budapest  2004

    DOI

  • Neural Networks with Branch Gates

    K. Goto, K. Hirasawa, J. Hu

    IJCNN 2004   Budapest  2004

    DOI

  • Stability Analysis of a DC Motor System Using Universal Learning Networks

    A. Hussein, K. Hirasawa, J. Hu

    IJCNN 2004   Budapest  2004

    DOI

  • Robust Feedback Error Learning Method for Controller Design of Nonlinear Systems

    H. Chen, K. Hirasawa, J. Hu

    IJCNN 2004   Budapest  2004

    DOI

  • A stock Prediction Model by Using Genetic Network Programming

    S. Mori, K. Hirasawa, J. Hu

    SICE 2004   Sapporo   1186 - 1191  2004

  • Genetic Network Programming with Reinforcement Learning for Generating Agent Behavior in the Benchmark Program

    S. Mabu, K. Hirasawa, J. Hu

    SICE 2004   Sapporo   918 - 923  2004

  • Studying the stability of a robust PV-supplied DC motor by universal learning networks

    A Hussein, K Hirasawa, JL Hu

    SICE 2004 ANNUAL CONFERENCE, VOLS 1-3   Sapporo   317 - 322  2004

     View Summary

    In this paper, a new robust control method and stability analysis method using the higher order derivatives (HODs) of universal learning networks (ULNs) is discussed. The application of these two methods to a Photovoltaic (PV) supplied, separately-excited DC motor with a constant load torque is also studied. The simulation results of the proposed robust control and stability analysis method showed that the robustness of the DC motor system is improved and its stability can be analyzed easily.

  • Functional localization of genetic network programming and its application to a dynamic problem

    S Eto, K Hirasawa, J Hu

    SICE 2004 ANNUAL CONFERENCE, VOLS 1-3   Sapporo   609 - 613  2004

     View Summary

    According to the knowledge of brain science, it is suggested that there exists cerebral functional localization, which means that a specific part of the cerebrum is activated depending on various kinds of information human receives. The aim of this paper is to build an artificial model to realize functional localization based on Genetic Network Programming (GNP), a new evolutionary computation method recently developed. GNP has a directed graph structure suitable for realizing functional localization. We studied the basic characteristics of the proposed system by making GNP work in a functionally localized way.

  • Training quasi-ARX neural network model by homotopy approach

    JL Hu, XB Lu, K Hirasawa

    SICE 2004 ANNUAL CONFERENCE, VOLS 1-3   Sapporo   367 - 372  2004

     View Summary

    Quasi-ARX neural networks (NN) are NN based nonlinear models that not only have linear structures similar to linear ARX models, but also have useful interpretation in part of their parameters. However when applying an ordinary backpropagation (BP) for the training, it has potential risk that the BP algorithm is stuck at a local minimum, which results in a poorly trained model. In this paper, a homotopy continuation method is introduced to improve the convergence performance of BP training. The idea is to start the BP training with the criterion function for linear ARX model, which is gradually deformed first into one for quasi-ARX NN model with linear node functions, and then into the actual one for quasi-ARX NN with sigmoid node functions. By building the deformation into a usual recursive procedure for BP training of quasi-ARX NN model with adaptable node functions so that the proposed homotopy based BP algorithm is able to achieve improved convergence performance without much increase in the computation load. Numerical simulation results show that the proposed homotopy based BP has better performance than an ordinary BP.

  • 共生と進化現象を統合する生態系のモデル化の研究

    山下貴志, 平澤宏太郎, 古月敬之, 武居雅暁

    Journal of Signal Processing   8 ( 1 ) 63 - 72  2004

    DOI CiNii

  • ネットワーク型アセンブリ言語を用いた人工生態系モデルの基礎検討

    白石優旗, 平澤宏太郎, 古月敬之, 村田純一

    電気学会論文誌C   124 ( 2 ) 418 - 424  2004

    DOI CiNii

  • マルチエージェントシステムの共生進化モデルの構築

    江口徹, 平澤宏太郎, 古月敬之

    情報処理学会論文誌   45 ( SIG2 ) 144 - 156  2004

  • 一般化学習ネットワークを利用した非線形離散時間動的システムの安定解析法

    平澤宏太郎, 古月敬之, ゆう, 間普真吾

    Journal of Signal Processing   8 ( 3 ) 235 - 247  2004

    DOI CiNii

  • 適応的離散ランダム探索法RasID-Dと最適化問題への適用

    平澤宏太郎, 宮崎弘幸, 古月敬之, 後藤健一

    Journal of Signal Processing   8 ( 4 ) 351 - 358  2004

    DOI CiNii

  • Genetic Network Programming with Automatically Generated Macro Nodes

    NAKAGOE Hiroshi, MABU Shingo, HIRASAWA Kotaro, HURUTSUKI Takayuki

    IEEJ Transactions on Electronics, Information and Systems   124 ( 8 ) 1619 - 1625  2004

     View Summary

    Genetic Network Programming (GNP) extended from other evolutionary computations such as Genetic Algorithm (GA) and Genetic Programming (GP) has network structures as gene. Previously, the program size of conventional GNP was fixed and GNP programs have not introduced the concept of sub-routines, although GA and GP paid attention to sub-routines. In this paper, a new method where GNP with Automatically Generated Macro Nodes (GNP with AGMs) composed of a number of nodes is proposed for improving the performance of GNP. These AGMs also have network structures and are evolved like main GNP. In addition to that, AGMs have multiple inputs and outputs that have not been introduced in the past. In the simulations, comparisons between GNP program only and GNP with AGMs are carried out using the tile world. Simulation results shows that the proposed method brings better results compared with traditional GNP. And it is clarified from simulations that the node transition rules obtained by AGMs show the generalized rules able to deal with unknown environments.

    DOI CiNii

  • Genetic Network Programming with Evolution and Learning and Its Application to the Tileworld Problem

    MABU Shingo, HIRASAWA Kotaro, HU Jinglu

    Transactions of the Society of Instrument and Control Engineers   40 ( 11 ) 1106 - 1113  2004

     View Summary

    A new evolutionary algorithm named "Genetic Network Programming, GNP" has been proposed. GNP represents its solutions as directed graphs, which realizes better expression ability than GA and GP which use string and tree structures, respectively. The aim of developing GNP is to deal with dynamic environments efficiently by using the distinguished expression ability and the inherently equipped functions of the network structure. However, since GNP is based on evolution, the programs cannot be changed until one generation ends. In this paper, we propose the extended algorithm, "GNP with Evolution and Learning" which combines evolution and reinforcement learning in order to adapt to dynamic environments quickly. The tileworld is used as a simulation environment and the results show some advantages of the proposed method.

    DOI CiNii

  • Associative Memory Constructed by Learning of Universal Learning Networks

    HIRASAWA Kotaro, SHIBUTA Keiko, FURUZUKI Takayuki, OTA Noriko

    IEEJ Transactions on Electronics, Information and Systems   124 ( 11 ) 2359 - 2367  2004

     View Summary

    Since the first neuron model was proposed, a lot of Neural Networks have been devised and been put into lots of practical uses. It is also true in the field of associative memory. Although so many useful memory models have been devised, there are still some problems, such as the limitation of storage capacity or too small attractor size to be stored.<br>In this paper, to solve the above problems, a novel associative memory is proposed. Its unique features are, (1) the memory network is obtained by training network parameters, (2) the size of the attractor of each stored memory can be controlled, and (3) some redundant nodes are introduced into the memory network in order to increase the storage capacity.<br>It is clarified from simulations that the proposed method can improve the memory functions, and can be applicable to the mutual associative memory easily.

    DOI CiNii

  • A robust control method for a PV-supplied DC motor using universal learning networks

    A Hussein, K Hirasawa, JL Hu

    SOLAR ENERGY   76 ( 6 ) 771 - 780  2004

     View Summary

    In this paper, a new robust control method and its application to a photovoltaic (PV) supplied, separately excited DC motor loaded with a constant torque is discussed. The robust controller is designed against the load torque changes by using the first and second ordered derivatives of the universal learning networks (ULNs). These derivatives are calculated using the forward propagation algorithm, which is considered as an extended version of real time recurrent learning (RTRL). In this application, two ULN&apos;s are used: The first is the ULN identifier trained offline to emulate the dynamic performance of the DC motor system. The second is the ULN controller, which is trained online not only to make the motor speed follow a selected reference signal, but also to make the overall system operate at the maximum power point of the PV source. To investigate the effectiveness of the proposed robust control method, the simulation is carried out at four different values of the robustness coefficient in two different stages: The training stage, in which the simulation is done for a constant load torque. And the control stage, in which the controller performance is obtained when the load torque is changed. The simulation results showed that the robustness of the control system is improved although the motor load torque at the control stage is different from that at the training stage. (C) 2004 Elsevier Ltd. All rights reserved.

    DOI CiNii

  • An Approximate Stability Analysis of a Robust Control DC Motor System

    A. Hussein, K. Hirasawa, J. Hu, K. Wada

    AROB 2003   Oita   333 - 336  2004

  • Evolution of Hybrid Neural Networks Using Genetic Network Programming

    D. Li, K. Hirasawa, J. Hu

    AROB 2003   Oita  2004

  • Genetic algorithm optimization of a convolutional neural network for autonomous crack detection

    R Ouellette, M Browne, K Hirasawa

    CEC2004: PROCEEDINGS OF THE 2004 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1 AND 2   Portland   516 - 521  2004

     View Summary

    Detecting cracks is an important function in building, tunnel, and bridge structural analysis. Successful automation of crack detection can provide a uniform and timely means for preventing further damage to structures. This laboratory has successfully applied convolutional neural networks (CNNs) to online crack detection. CNNs represent an interesting method for adaptive image processing, and form a link between artificial neural networks and finite impulse response filters. As with most artificial neural networks, the CNN is susceptible to multiple local minima, thus complexity and time must be applied in order to avoid becoming trapped within the local minima. This paper employs a standard genetic algorithm (GA) to train the weights of a 4-5x5 filter CNN in order to pass through local minima. This technique resulted in a 92.3 +/- 1.4% average success rate using 25 GA-trained CNNs presented with 100 crack (320x240 pixel) images.

  • Elevator group supervisory control systems using genetic network programming

    T Eguchi, K Hirasawa, J Hu, S Markon

    CEC2004: PROCEEDINGS OF THE 2004 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1 AND 2   Portland   1661 - 1667  2004

     View Summary

    Genetic Network Programming (GNP) has been proposed as a new method of evolutionary computation. Until now, GNP has been applied to various problems and its effectiveness was clarified. However, these problems were virtual models, so the applicability and availability of GNP to the real-world applications have not been studied. In this paper, as a first step of applying GNP to the real-world applications, Elevator Group Supervisory Control Systems (EGSCSs) are considered. Generally, EGSCSs are complex and difficult problems to solve because they are too dynamic and probabilistic. So the design of a useful controller of EGSCSs was very difficult. Recently, the design of such a controller of EGSCSs has been tried actively using Artificial Intelligence (AI) technologies. In this paper, it is reported that the design of a controller of EGSCSs has been studied using GNP whose characteristic is to use directed graph as its gene instead of bit strings and trees of GA and GP From simulations, it is clarified that better solutions are obtained by using GNP than other conventional methods and the availability of GNP to real-world applications is confirmed.

  • Genetic network programming with automatically generated variable size macro nodes

    H Nakagoe, K Hirasawa, J Hu

    CEC2004: PROCEEDINGS OF THE 2004 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1 AND 2   Portland   713 - 719  2004

     View Summary

    Genetic Network Programming (GNP) has directed graph structures as genes, which is extended from other evolutionary computations such as Genetic Algorithm (GA) and Genetic Programming (GP). Generally, macroinstructions are introduced as sub-routines, function localization and so on. Previously, we have introduced the structure of macroinstructions in GNP named Automatically Generated Macro Nodes (AGMs) for reducing the time of evolution efficiently, and showed that macroinstructions are useful to acquire good performances. But the AGMs have fixed number of nodes, and it is found that the effectiveness of evolution of macroinstructions depends on the main program calling them and initialized parameters. Accordingly in this paper, new AGMs are introduced to improve their performances further more by the mechanism of varying the size of AGMs, which are named variable size AGMs. This is the mechanism to add and delete nodes according to necessity. In the simulations, comparisons between GNP program only, GNP with conventional AGMs and GNP with variable size AGMs are carried out using the tile world. Simulation results show that the proposed method is better compared with conventional GNP and GNP with conventional AGMs. And also it is clarified that the node transition rules obtained by new AGMs show the generalized rules able to deal with unknown environments.

  • Functional localization of genetic network programming and its application to a pursuit problem

    SJ Eto, K Hirasawa, J Hu

    CEC2004: PROCEEDINGS OF THE 2004 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1 AND 2   Portland   683 - 690  2004

     View Summary

    According to the knowledge of brain science, it is suggested that there exists cerebral functional localization, which means that a specific part of the cerebrum is activated depending on various kinds of information human receives. The aim of this paper is to build an artificial model to realize functional localization based on Genetic Network Programming (GNP), a new evolutionary computation method recently developed. GNP has a directed graph structure suitable for realizing functional localization. We studied the basic characteristics of the proposed system by making GNP work in a functionally localized way.

  • Genetic Network Programming with Reinforcement Learning and its Performance Evaluation

    S. Mabu, K. Hirasawa, J. Hu

    GECCO 2004   Seattle  2004

  • Self-Organized Functional Localization Neural Network

    T. Sasakawa, J. Hu, K. Hirasawa

    IJCNN 2004   Budapest  2004

    DOI

  • Multi-Branch Structure and Its Localized Property in Layered Neural Networks

    T. Yamasita, K. Hirasawa, J. Hu

    IJCNN 2004   Budapest  2004

    DOI

  • Neural Networks with Branch Gates

    K. Goto, K. Hirasawa, J. Hu

    IJCNN 2004   Budapest  2004

    DOI

  • Stability Analysis of a DC Motor System Using Universal Learning Networks

    A. Hussein, K. Hirasawa, J. Hu

    IJCNN 2004   Budapest  2004

    DOI

  • Robust Feedback Error Learning Method for Controller Design of Nonlinear Systems

    H. Chen, K. Hirasawa, J. Hu

    IJCNN 2004   Budapest  2004

    DOI

  • A stock Prediction Model by Using Genetic Network Programming

    S. Mori, K. Hirasawa, J. Hu

    SICE 2004   Sapporo   1186 - 1191  2004

  • Genetic Network Programming with Reinforcement Learning for Generating Agent Behavior in the Benchmark Program

    S. Mabu, K. Hirasawa, J. Hu

    SICE 2004   Sapporo   918 - 923  2004

  • Studying the stability of a robust PV-supplied DC motor by universal learning networks

    A Hussein, K Hirasawa, JL Hu

    SICE 2004 ANNUAL CONFERENCE, VOLS 1-3   Sapporo   317 - 322  2004

     View Summary

    In this paper, a new robust control method and stability analysis method using the higher order derivatives (HODs) of universal learning networks (ULNs) is discussed. The application of these two methods to a Photovoltaic (PV) supplied, separately-excited DC motor with a constant load torque is also studied. The simulation results of the proposed robust control and stability analysis method showed that the robustness of the DC motor system is improved and its stability can be analyzed easily.

  • Functional localization of genetic network programming and its application to a dynamic problem

    S Eto, K Hirasawa, J Hu

    SICE 2004 ANNUAL CONFERENCE, VOLS 1-3   Sapporo   609 - 613  2004

     View Summary

    According to the knowledge of brain science, it is suggested that there exists cerebral functional localization, which means that a specific part of the cerebrum is activated depending on various kinds of information human receives. The aim of this paper is to build an artificial model to realize functional localization based on Genetic Network Programming (GNP), a new evolutionary computation method recently developed. GNP has a directed graph structure suitable for realizing functional localization. We studied the basic characteristics of the proposed system by making GNP work in a functionally localized way.

  • Training quasi-ARX neural network model by homotopy approach

    JL Hu, XB Lu, K Hirasawa

    SICE 2004 ANNUAL CONFERENCE, VOLS 1-3   Sapporo   367 - 372  2004

     View Summary

    Quasi-ARX neural networks (NN) are NN based nonlinear models that not only have linear structures similar to linear ARX models, but also have useful interpretation in part of their parameters. However when applying an ordinary backpropagation (BP) for the training, it has potential risk that the BP algorithm is stuck at a local minimum, which results in a poorly trained model. In this paper, a homotopy continuation method is introduced to improve the convergence performance of BP training. The idea is to start the BP training with the criterion function for linear ARX model, which is gradually deformed first into one for quasi-ARX NN model with linear node functions, and then into the actual one for quasi-ARX NN with sigmoid node functions. By building the deformation into a usual recursive procedure for BP training of quasi-ARX NN model with adaptable node functions so that the proposed homotopy based BP algorithm is able to achieve improved convergence performance without much increase in the computation load. Numerical simulation results show that the proposed homotopy based BP has better performance than an ordinary BP.

  • A robust control method for a PV-supplied DC motor using universal learning networks

    A Hussein, K Hirasawa, JL Hu

    SOLAR ENERGY   76 ( 6 ) 771 - 780  2004

     View Summary

    In this paper, a new robust control method and its application to a photovoltaic (PV) supplied, separately excited DC motor loaded with a constant torque is discussed. The robust controller is designed against the load torque changes by using the first and second ordered derivatives of the universal learning networks (ULNs). These derivatives are calculated using the forward propagation algorithm, which is considered as an extended version of real time recurrent learning (RTRL). In this application, two ULN&apos;s are used: The first is the ULN identifier trained offline to emulate the dynamic performance of the DC motor system. The second is the ULN controller, which is trained online not only to make the motor speed follow a selected reference signal, but also to make the overall system operate at the maximum power point of the PV source. To investigate the effectiveness of the proposed robust control method, the simulation is carried out at four different values of the robustness coefficient in two different stages: The training stage, in which the simulation is done for a constant load torque. And the control stage, in which the controller performance is obtained when the load torque is changed. The simulation results showed that the robustness of the control system is improved although the motor load torque at the control stage is different from that at the training stage. (C) 2004 Elsevier Ltd. All rights reserved.

    DOI CiNii

  • 学習・進化型遺伝的ネットワークプログラミングとそのマルチエージェントシステムへの応用

    間普真吾, 平澤宏太郎, 胡敬炉

    情報処理学会進化型計算シンポジウム   京都  2003

  • タイルワールドモデルを用いた共生学習進化型マルチエージェントシステム

    江口徹, 平澤宏太郎, 胡敬炉

    情報処理学会進化型計算シンポジウム   京都  2003

  • Neural Networks with Adaptive Node Function

    後藤健一, 平澤宏太郎, 古月敬之

    SICEシステム情報部門学術講演会   仙台   313 - 316  2003

  • Task-Oriented Multiagent Reinforcement Learning Control for a Real Time High-Dimensional Problem

    KAMAL M. A. S.

    Proc. of the eighth International Symposium on Artificial Life and Robotics, 2003   Beppu   353 - 356  2003

    CiNii

  • 2D Artificial Life System Using Network-type Assembly-like Language: A Comparative Study with Linear-type Assembly-like Language

    Y. Shiraishi, K. Hirasawa, J. Hu, J. Murata

    AROB 2002   Beppu   669 - 672  2003

  • Online Identification and Control of A PV-Supplied DC Motor Using Universal Learning Networks

    A. Hussein, K. Hirasawa, J. Hu

    ESANN 2003   Belgium  2003

  • An Application of Universal Learning Networks in the Speed Control of DC Motor Drives Fed from a Photovoltaic Generator

    A. Hussein, K. Hirasawa, J. Hu

    DCDIS 2003   Guelph  2003

  • An Efficient Constructive Higher Order Neural Network with Multiplication Units

    D. Li, K. Hirasawa, J. Hu

    ICANN/ICONIP 2003   Istanbul   21 - 24  2003

  • An Adaptive Speed Controller for A PV-Supplied DC Motor Using Universal Learning Networks

    A. Hussein, K. Hirasawa, J. Hu

    ICANN/ICONIP 2003   Istanbul   5 - 8  2003

  • Genetic Network Programming with Automatically Generated Macro Nodes

    H. Nakagoe, S. Mabu, K. Hirasawa, J. Hu

    SICE 2003   Fukui   1486 - 1491  2003

  • Evolutional Acquisition of Communication of GNP

    S. Eto, T. Eguchi, K. Hirasawa, J. Hu

    SICE 2003   Fukui   1215 - 1220  2003

  • Universal Learning Networks with Adaptive Node Functions

    K. Goto, K. Hirasawa, J. Hu

    SICE 2003   Fukui   1492 - 1497  2003

  • A New Strategy for Constructing Higher Order Neural Networks with Multiplication Units

    D. Li, K. Hirasawa, J. Hu

    SICE 2003   Fukui   606 - 611  2003

  • Multi-Branch Neural Networks with Branch Control

    T. yamasita, K. Hirasawa, J. Hu

    SICE 2003   Fukui   1784 - 1789  2003

  • Robust Control of a PV-Supplied DC Motor Using Higher Order Derivatives of ULNs

    A. Hussein, K. Hirasawa, J. Hu

    SICE 2003   Fukui   26 - 31  2003

  • Identification and Control of a PV-Supplied Separately Excited DC Motor Using Universal Learning Networks

    A. Hussein, K. Hirasawa, J. Hu

    IFAC Symp. on Identification 2003   Rotterdam   1454 - 1459  2003

  • Multi-Branch Neural Networks with Branch Control

    T. Yamasita, K. Hirasawa, J. Hu

    IEEE SMC 2003   Washington DC   756 - 761  2003

  • Symbiotic evolutional models in multiagent systems

    T Eguchi, K Hirasawa, J Hu

    CEC: 2003 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1-4, PROCEEDINGS   Canberra   739 - 746  2003

     View Summary

    Multiagent Systems with Symbiotic Learning and Evolution (Masbiole) has been proposed as a new learning and evolutionary method for Multiagent Systems (MAS) recently, which is based on symbiotic phenomena among creatures. In this paper, a symbiotic evolutional model of Masbiole is proposed using Genetic Network Programming (GNP), which has been also proposed as one of the evolutionary computations. In the simulations, the proposal Masbiole is applied to the tile-world model and various characteristics of Masbiole have been clarified.

    DOI

  • Genetic network programming with learning and evolution for adapting to dynamical environments

    S Mabu, K Hirasawa, JL Hu

    CEC: 2003 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1-4, PROCEEDINGS   Canberra   69 - 76  2003

     View Summary

    A new evolutionary algorithm named "Genetic Network Programming, GNP" has been proposed. GNP represents its solutions as network structures, which can improve the expression and search ability. Since GA, GP and GNP already proposed are based on evolution and they cannot change their solutions until one generation ends, we propose GNP with Learning and Evolution in order to adapt to dynamical environments quickly. Learning algorithm improves search speed for solutions and Evolutionary algorithm enables GNP to search wide solution space efficiently.

    DOI

  • Multi Agent Systems with Symbiotic Learning and Evolution -Masbiole- and Its Application

    HIRASAWA Kotaro, NAKANISHI Katsushige, EGUCHI Toru, HU Jinglu

    IEEJ Transactions on Electronics, Information and Systems   123 ( 1 ) 67 - 74  2003

     View Summary

    Recently, systems are becoming more complex and larger than ever, so numerous attempts have been made to introduce biological features into artificial systems, because many biological systems in the nature exist as one of the most complex systems.<br>Multi agent system with symbiotic learning and evolution have been recently proposed. It is named Masbiole. In this paper, Masbiole is reviewed and the method for evolving multi agent systems is proposed. From simulations on a multi objective knapsack problem, it has been clarified that Masbiole has better performance than that of conventional multi objective genetic algorithms.

    DOI CiNii

  • Variable Size Genetic Network Programming

    Katagiri Hironobu, Hu Jinglu, Murata Junichi, Hirasawa Kotaro

    IEEJ Transactions on Electronics, Information and Systems   123 ( 1 ) 57 - 66  2003

     View Summary

    Genetic Network Programming (GNP) is a kind of evolutionary methods, which evolves arbitrary directed graph programs. Previously, the program size of GNP was fixed. In the paper, a new method is proposed, where the program size is adaptively changed depending on the frequency of the use of nodes. To control and to decide a program size are important and difficult problems in Evolutionary Computation, especially, a well-known crossover operator tends to cause bloat. We introduce two additional operators, add operator and delete operator, that can change the number of each kind of nodes based on whether a node function is important in the environment or not. Simulation results shows that the proposed method brings about extremely better results compared with ordinary fixed size GNP. © 2003, The Institute of Electrical Engineers of Japan. All rights reserved.

    DOI CiNii

  • ノード数減少方RBFネットワークとその制御器設計問題への応用

    伊藤信治, 村田純一, 平澤宏太郎

    電気学会論文誌C   123 ( 2 ) 338 - 344  2003

    DOI CiNii

  • Neural Network Based Prediction Model for Control of Nonlinear Systems

    HU Jinglu, HIRASAWA Kotaro

    Transactions of the Society of Instrument and Control Engineers   39 ( 2 ) 168 - 175  2003

     View Summary

    Neural networks have attracted much interest in system identification and control communities because they can learn any nonlinear mapping. However, from a user's point of view, when neural networks are used as models for controller design, they do not have structures of easy use. This paper introduces a new neural network based prediction model for control of nonlinear systems. Distinctive features of the new model to conventional neural-network based ones are that it has not only meaningful interpretation on part of its parameters but also is linear for the input variables. The former feature makes parameter estimation easier and the latter allows us to derive a nonlinear controller directly from the identified prediction model. The modeling and the parameter estimation are described in detail. The usefulness of the new model is demonstrated by using numerical simulations.

    DOI CiNii

  • Hybrid Universal Learning Networks

    Dazi Li, Jinglu Hu, Junichi Murata, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   123 ( 3 ) 552 - 559  2003

     View Summary

    A variety of neuron models combine the neural inputs through their summation and sigmoidal functions. Such structure of neural networks leads to shortcomings such as a large number of neurons in hidden layers and huge training data required. We introduce a kind of multiplication neuron which multiplies their inputs instead of summing to overcome the above problems. A hybrid universal learning network constructed by the combination of multiplication units arid summation units is proposed and trained for several well known benchmark problems. Different combinations of the above two are tried. It is clarified that multiplication is an essential computational element in many cases and the combination of the multiplication units with summation units in different layers in the networks improved the performance of the network. © 2003, The Institute of Electrical Engineers of Japan. All rights reserved.

    DOI CiNii

  • Online Learning of Genetic Network Programming and its Application to Prisoner's Dilemma Game

    Shingo Mabu, Jinglu Hu, Junichi Murata, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   123 ( 3 ) 535 - 543  2003

     View Summary

    A new evolutionary model with the network structure named Genetic Network Programming (GNP) has been proposed recently. GNP, that is, an expansion of GA and GP, represents solutions as a network structure and evolves it by using “offline learning (selection, mutation, crossover)”. GNP can memorize the past action sequences in the network flow, so it can deal with Partially Observable Markov Decision Process (POMDP) well. In this paper, in order to improve the ability of GNP, Q learning (an off-policy TD control algorithm) that is one of the famous online methods is introduced for online learning of GNP. Q learning is suitable for GNP because (1) in reinforcement learning, the rewards an agent will get in the future can be estimated, (2) TD control doesn't need much memory and can learn quickly, and (3) off-policy is suitable in order to search for an optimal solution independently of the policy. Finally, in the simulations, online learning of GNP is applied to a player for “Prisoner's dilemma game“ and its ability for online adaptation is confirmed. © 2003, The Institute of Electrical Engineers of Japan. All rights reserved.

    DOI CiNii

  • Genetic Network Programmingを用いた共生学習進化型マルチエージェントシステム

    江口徹, 平澤宏太郎, 胡敬炉, 村田純一

    電気学会論文誌C   123 ( 3 ) 517 - 526  2003

    DOI CiNii

  • Genetic Network Programmingによるヘテロマルチエージェントシステムの構成

    平澤宏太郎, 大久保雅文, 胡敬炉, 村田純一, 松家裕子

    電気学会論文誌C   123 ( 3 ) 544 - 551  2003

    DOI CiNii

  • Function Approximation Using LVQ

    KYU Shon MIN, MURATA Junichi, HIRASAWA Kotaro

    Transactions of the Society of Instrument and Control Engineers   39 ( 5 ) 513 - 519  2003

     View Summary

    Neural networks with local activation functions, for example RBFNs (Radial Basis Function Networks), have a merit of excellent generalization abilities. When this type of network is used in function approximation, it is very important to determine the proper division of the input space into local regions to each of which a local activation function is assigned. In RBFNs, this is equivalent to determination of the locations and the numbers of its RBFs, which is generally done based on the distribution of input data. But, in function approximation, the output information (the value of the function to be approximated) must be considered in determination of the local regions. A new method is proposed that uses LVQ network to approximate functions based on the output information. It divides the input space into regions with a prototype vector at the center of each region. The ordinary LVQ, however, outputs discrete values only, and therefore can not deal with continuous functions. In this paper, a technique is proposed to solve this problem. Examples are provided to show the effectiveness of the proposed method.

    DOI CiNii

  • Enhancing the Generalization Ability of Neural Networks by Using Gram-Schmidt Orthogonalization Algorithm

    WAN Weishui, HIRASAWA Kotaro, HU Jinglu

    Transactions of the Society of Instrument and Control Engineers   39 ( 7 ) 697 - 698  2003

     View Summary

    In this paper a new algorithm applying Gram-Schmidt orthogonalization algorithm to the outputs of nodes in the hidden layers is proposed with the aim to reduce the interference among the nodes in the hidden layers, therefore to enhance the generalization ability of neural networks, which is much more efficient than other regularizers methods. Simulation results confirm the above assertion.

    DOI CiNii

  • Increasing Robustness of Binary-coded Genetic Algorithm

    Jiangming Mao, Junichi Murata, Kotaro Hirasawa, Jinglu Hu

    IEEJ Transactions on Electronics, Information and Systems   123 ( 9 ) 1625 - 1630  2003

     View Summary

    Genetic algorithms are often well suited for optimization problems because of their parallel searching and evolutionary ability. Crossover and mutation are believed to be the main exploration operators. In this paper, we focus on how crossover and mutation work in binary-coded genetic algorithm and investigate their effects on bit's frequency of population. According to the analysis of equilibrium of crossover, we can see the bit-based simulated crossover (BSC) is strong crossover method. Furthermore, to increase robustness of binary-coded genetic algorithm, multi-generation inheritance evolutionary strategy(MGIS) was proposed. Simulation results demonstrate the effectiveness of the proposed method. © 2003, The Institute of Electrical Engineers of Japan. All rights reserved.

    DOI CiNii

  • Neural Networks with Node Gates which Divide and Conquer Problems Based on Local Difficulties

    MURATA Junichi, KAJIWARA Yoshitatsu, HIRASAWA Kotaro

    Transactions of the Society of Instrument and Control Engineers   39 ( 9 ) 841 - 847  2003

     View Summary

    A neural network is proposed based on a divide-and-conquer scheme. The network has gates which control firing of its hidden nodes. By opening and closing the gates depending on input values, the network divides the input space into sub-regions and assigns its nodes to each of them to produce the desired output in that region. The division mechanism is constructed by learning. A new learning method is proposed which divides the space in accordance to the difficulties; areas with larger errors are divided into smaller sub-regions. Thus, the nodes in the network are more densely assigned to areas with higher difficulties to 'conquer' the areas appropriately. Function approximation examples are provided to illustrate the validity of the proposed network.

    DOI CiNii

  • パラメータ可変一般化学習ネットワークの理論検討

    平澤宏太郎, 山下貴志, 古月敬之, 李大字

    Journal of Signal Processing   7 ( 6 ) 411 - 420  2003

  • Multi-Branch Structure of Layered Neural Networks

    T. Yamasita, K. Hirasawa, J. Hu

    J. Machine Intelligence and Robotic Control   5 ( 1 ) 17 - 23  2003

  • Evolutional Acquisition of Communication Between Agents Using Genetic Network Programming

    S. Eto, T. Eguchi, K. Hirasawa, J. Hu

    J. Machine Intelligence and Robotic Control   5 ( 1 ) 25 - 32  2003

  • Multiple Probability Vectors Based Genetic Algorithm

    J. Mao, K. Hirasawa, J. Hu

    J. Machine Intelligence and Robotoc Control   5 ( 2 ) 59 - 67  2003

  • A Robust Controller Design Method of Nonlinear Systems Based on Feedback Error Learning

    H. Chen, K. Hirasawa, J. Hu

    J. Machine Intelligence and Robot Control   5 ( 4 ) 121 - 128  2003

  • A Function Localized Neural Network with Branch Control

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata

    Neural Networks   16 ( 10 ) 1461 - 1481  2003

    DOI PubMed

  • Neural Networks with Adaptive Node Function

    後藤健一, 平澤宏太郎, 古月敬之

    SICEシステム情報部門学術講演会   仙台   313 - 316  2003

  • Task-Oriented Multiagent Reinforcement Learning Control for a Real Time High-Dimensional Problem

    S. Kamal, J. Murata, K. Hirasawa

    AROB 2002   Beppu   353 - 356  2003

  • 2D Artificial Life System Using Network-type Assembly-like Language: A Comparative Study with Linear-type Assembly-like Language

    Y. Shiraishi, K. Hirasawa, J. Hu, J. Murata

    AROB 2002   Beppu   669 - 672  2003

  • Online Identification and Control of A PV-Supplied DC Motor Using Universal Learning Networks

    A. Hussein, K. Hirasawa, J. Hu

    ESANN 2003   Belgium  2003

  • An Application of Universal Learning Networks in the Speed Control of DC Motor Drives Fed from a Photovoltaic Generator

    A. Hussein, K. Hirasawa, J. Hu

    DCDIS 2003   Guelph  2003

  • An Efficient Constructive Higher Order Neural Network with Multiplication Units

    D. Li, K. Hirasawa, J. Hu

    ICANN/ICONIP 2003   Istanbul   21 - 24  2003

  • An Adaptive Speed Controller for A PV-Supplied DC Motor Using Universal Learning Networks

    A. Hussein, K. Hirasawa, J. Hu

    ICANN/ICONIP 2003   Istanbul   5 - 8  2003

  • Genetic Network Programming with Automatically Generated Macro Nodes

    H. Nakagoe, S. Mabu, K. Hirasawa, J. Hu

    SICE 2003   Fukui   1486 - 1491  2003

  • Evolutional Acquisition of Communication of GNP

    S. Eto, T. Eguchi, K. Hirasawa, J. Hu

    SICE 2003   Fukui   1215 - 1220  2003

  • Universal Learning Networks with Adaptive Node Functions

    K. Goto, K. Hirasawa, J. Hu

    SICE 2003   Fukui   1492 - 1497  2003

  • A New Strategy for Constructing Higher Order Neural Networks with Multiplication Units

    D. Li, K. Hirasawa, J. Hu

    SICE 2003   Fukui   606 - 611  2003

  • Multi-Branch Neural Networks with Branch Control

    T. yamasita, K. Hirasawa, J. Hu

    SICE 2003   Fukui   1784 - 1789  2003

  • Robust Control of a PV-Supplied DC Motor Using Higher Order Derivatives of ULNs

    A. Hussein, K. Hirasawa, J. Hu

    SICE 2003   Fukui   26 - 31  2003

  • Identification and Control of a PV-Supplied Separately Excited DC Motor Using Universal Learning Networks

    A. Hussein, K. Hirasawa, J. Hu

    IFAC Symp. on Identification 2003   Rotterdam   1454 - 1459  2003

  • Multi-Branch Neural Networks with Branch Control

    T. Yamasita, K. Hirasawa, J. Hu

    IEEE SMC 2003   Washington DC   756 - 761  2003

  • Symbiotic evolutional models in multiagent systems

    T Eguchi, K Hirasawa, J Hu

    CEC: 2003 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1-4, PROCEEDINGS   Canberra   739 - 746  2003

     View Summary

    Multiagent Systems with Symbiotic Learning and Evolution (Masbiole) has been proposed as a new learning and evolutionary method for Multiagent Systems (MAS) recently, which is based on symbiotic phenomena among creatures. In this paper, a symbiotic evolutional model of Masbiole is proposed using Genetic Network Programming (GNP), which has been also proposed as one of the evolutionary computations. In the simulations, the proposal Masbiole is applied to the tile-world model and various characteristics of Masbiole have been clarified.

    DOI

  • Genetic network programming with learning and evolution for adapting to dynamical environments

    S Mabu, K Hirasawa, JL Hu

    CEC: 2003 CONGRESS ON EVOLUTIONARY COMPUTATION, VOLS 1-4, PROCEEDINGS   Canberra   69 - 76  2003

     View Summary

    A new evolutionary algorithm named "Genetic Network Programming, GNP" has been proposed. GNP represents its solutions as network structures, which can improve the expression and search ability. Since GA, GP and GNP already proposed are based on evolution and they cannot change their solutions until one generation ends, we propose GNP with Learning and Evolution in order to adapt to dynamical environments quickly. Learning algorithm improves search speed for solutions and Evolutionary algorithm enables GNP to search wide solution space efficiently.

    DOI

  • Hybrid Universal Learning Networks

    Dazi Li, Jinglu Hu, Junichi Murata, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   123 ( 3 ) 552 - 559  2003

     View Summary

    A variety of neuron models combine the neural inputs through their summation and sigmoidal functions. Such structure of neural networks leads to shortcomings such as a large number of neurons in hidden layers and huge training data required. We introduce a kind of multiplication neuron which multiplies their inputs instead of summing to overcome the above problems. A hybrid universal learning network constructed by the combination of multiplication units arid summation units is proposed and trained for several well known benchmark problems. Different combinations of the above two are tried. It is clarified that multiplication is an essential computational element in many cases and the combination of the multiplication units with summation units in different layers in the networks improved the performance of the network. © 2003, The Institute of Electrical Engineers of Japan. All rights reserved.

    DOI CiNii

  • Online Learning of Genetic Network Programming and its Application to Prisoner's Dilemma Game

    Shingo Mabu, Jinglu Hu, Junichi Murata, Kotaro Hirasawa

    IEEJ Transactions on Electronics, Information and Systems   123 ( 3 ) 535 - 543  2003

     View Summary

    A new evolutionary model with the network structure named Genetic Network Programming (GNP) has been proposed recently. GNP, that is, an expansion of GA and GP, represents solutions as a network structure and evolves it by using “offline learning (selection, mutation, crossover)”. GNP can memorize the past action sequences in the network flow, so it can deal with Partially Observable Markov Decision Process (POMDP) well. In this paper, in order to improve the ability of GNP, Q learning (an off-policy TD control algorithm) that is one of the famous online methods is introduced for online learning of GNP. Q learning is suitable for GNP because (1) in reinforcement learning, the rewards an agent will get in the future can be estimated, (2) TD control doesn't need much memory and can learn quickly, and (3) off-policy is suitable in order to search for an optimal solution independently of the policy. Finally, in the simulations, online learning of GNP is applied to a player for “Prisoner's dilemma game“ and its ability for online adaptation is confirmed. © 2003, The Institute of Electrical Engineers of Japan. All rights reserved.

    DOI CiNii

  • Function Approximation Using LVQ

    S. Min Kyu, J. Murata, K. Hirasawa

    計測自動制御学会論文集   39 ( 5 ) 513 - 519  2003

     View Summary

    Neural networks with local activation functions, for example RBFNs (Radial Basis Function Networks), have a merit of excellent generalization abilities. When this type of network is used in function approximation, it is very important to determine the proper division of the input space into local regions to each of which a local activation function is assigned. In RBFNs, this is equivalent to determination of the locations and the numbers of its RBFs, which is generally done based on the distribution of input data. But, in function approximation, the output information (the value of the function to be approximated) must be considered in determination of the local regions. A new method is proposed that uses LVQ network to approximate functions based on the output information. It divides the input space into regions with a prototype vector at the center of each region. The ordinary LVQ, however, outputs discrete values only, and therefore can not deal with continuous functions. In this paper, a technique is proposed to solve this problem. Examples are provided to show the effectiveness of the proposed method.

    DOI CiNii

  • Enhancing the Generalization Ability of Neural Networks by Using Gram-Schmidt Orthogonarization Algorithm

    W. Wan, K. Hirasawa, J. Hu

    計測自動制御学会論文集   39 ( 7 ) 697 - 698  2003

     View Summary

    In this paper a new algorithm applying Gram-Schmidt orthogonalization algorithm to the outputs of nodes in the hidden layers is proposed with the aim to reduce the interference among the nodes in the hidden layers, therefore to enhance the generalization ability of neural networks, which is much more efficient than other regularizers methods. Simulation results confirm the above assertion.

    DOI CiNii

  • Increasing Robustness of Binary-coded Genetic Algorithm

    Jiangming Mao, Junichi Murata, Kotaro Hirasawa, Jinglu Hu

    IEEJ Transactions on Electronics, Information and Systems   123 ( 9 ) 1625 - 1630  2003

     View Summary

    Genetic algorithms are often well suited for optimization problems because of their parallel searching and evolutionary ability. Crossover and mutation are believed to be the main exploration operators. In this paper, we focus on how crossover and mutation work in binary-coded genetic algorithm and investigate their effects on bit's frequency of population. According to the analysis of equilibrium of crossover, we can see the bit-based simulated crossover (BSC) is strong crossover method. Furthermore, to increase robustness of binary-coded genetic algorithm, multi-generation inheritance evolutionary strategy(MGIS) was proposed. Simulation results demonstrate the effectiveness of the proposed method. © 2003, The Institute of Electrical Engineers of Japan. All rights reserved.

    DOI CiNii

  • Multi-Branch Structure of Layered Neural Networks

    T. Yamasita, K. Hirasawa, J. Hu

    J. Machine Intelligence and Robotic Control   5 ( 1 ) 17 - 23  2003

  • Evolutional Acquisition of Communication Between Agents Using Genetic Network Programming

    S. Eto, T. Eguchi, K. Hirasawa, J. Hu

    J. Machine Intelligence and Robotic Control   5 ( 1 ) 25 - 32  2003

  • Multiple Probability Vectors Based Genetic Algorithm

    J. Mao, K. Hirasawa, J. Hu

    J. Machine Intelligence and Robotoc Control   5 ( 2 ) 59 - 67  2003

  • A Robust Controller Design Method of Nonlinear Systems Based on Feedback Error Learning

    H. Chen, K. Hirasawa, J. Hu

    J. Machine Intelligence and Robot Control   5 ( 4 ) 121 - 128  2003

  • A Function Localized Neural Network with Branch Control

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata

    Neural Networks   16 ( 10 ) 1461 - 1481  2003

    DOI PubMed

  • マルチブランチを用いたニューラルネットワーク

    山下貴志, 平澤宏太郎, 胡敬炉

    SICEシステム情報部門学術講演会   横浜   461 - 466  2002

  • 共生学習進化型マルチエージェントシステムとそのタイルワールドへの適用

    江口徹, 平澤宏太郎, 胡敬炉

    SICEシステム情報部門学術講演会   横浜   207 - 212  2002

  • ネットワーク型アセンブリ言語を用いた人工生態系モデルの基礎検討

    白石優旗, 平澤宏太郎, 胡敬炉, 村田純一

    FANシンポジウム   佐賀   119 - 122  2002

  • 学習・進化型遺伝的ネットワークプログラミング

    間普真吾, 平澤宏太郎, 胡敬炉, 村田純一

    FANシンポジウム   佐賀   1 - 6  2002

  • コミュニケーションを考慮したGNPによるマルチエージェントシステム

    江藤慎治, 平澤宏太郎, 胡敬炉, 江口徹

    SICE九州支部学術講演会   大分   269 - 272  2002

  • 一般化学習ネットワークによる連想記憶モデル

    渋田敬子, 胡敬炉, 平澤宏太郎

    SICE九州支部学術講演会   大分   249 - 252  2002

  • Online Learning of Genetic Network Programming(GNP)

    S. Mabu, K. Hirasawea, J. HU, J. Murata

    IEEE CEC 2002   Honolulu  2002

    DOI

  • Networks with Input Gates for Situation-Dependent Input Selection in Reinforecement Learning

    J. Murata, M.Suzuki, K. Hirasawa

    IJCNN 2002   Honolulu  2002

  • Learning of Symbiotic Relations among Agents by Using Neural Networks

    K. Hirasawa, H. Yoshida, K. Nakanishi

    IJCNN 2002   Honolulu  2002

  • The Dynamic Performance of Photovoltaic Supplied DC Motor Fed from DC-DC Converter by Neural Networks

    A. Hussein, K. Hirasawa, J. HU, J. Murata

    IJCNN 2002   Honolulu  2002

  • Training a Kind of Hybrid Universal Learning Networks with Classification Problems

    D. Li, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2002   Honolulu  2002

  • Adaptive Neural Network Speed Controller for DC Motor Drives

    A. Hussein, K. Hirasawa, J. Hu, J. Murata

    AdCONIP 2002   Kumamoto   465 - 470  2002

  • A Robust Controller Design Method of Nonlinear Systems for System Parameters Changes

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    AdCONIP 2002   Kumamoto   533 - 538  2002

  • A New Model to Realize Variable Size Genetic Network Programming-A Case Study with the Tileworld Problem

    H. Katagiri, K. Hirasawa, J. Hu, J. Murata

    GECCO 2002   New York   279 - 286  2002

  • Multiagent systems with symbiotic learning and evolution using Genetic Network Programming

    EGUCHI T.

    Late Breaking Papers at the Genetic and Evolutionary Computation Conference, 2002   New York   130 - 137  2002

    CiNii

  • The Basic Study of Artificial Ecosystem Models Using Network-Type Assembly-Like Language.

    Yuhki Shiraishi, Kotaro Hirasawa, Jinglu Hu, Junichi Murata

      New York   412 - 418  2002

  • Increasing Robustness of Genetic Algorithm

    J. Mao, K. Hoirasawa, J. Hu, J. Murata

    GECCO 2002   New York  2002

  • Auto Correlation Associative Memory by Universal Learning Networks

    K. Shibuta, K. Hirasawa, J. Hu, J. Murata

    SICE 2002   Osaka   793 - 798  2002

  • Multi-branch Structure of Layered Neural Networks

    T. Yamasita, K. Hirasawa, J. Hu, J. Murata

    SICE 2002   Osaka   799 - 804  2002

  • Task-Oriented Reiforcement Learning for Continuous Tasks in Dynamic Environment

    M. Kamal, J. Murata, K. Hirasawa

    SICE 2002   Osaka   932 - 935  2002

  • Comparing Some Graph Crossover in Genetic Network Programming

    H. Katagiri, K. Hirasawa, J. Hu, J. Murata

    SICE 2002   Osaka   936 - 941  2002

  • Automatic Generation of Programs Using Genetic Network Programming

    Y. Matsuya, K. Hirasawa, J. Hu, J. Murata

    SICE 2002   Osaka   942 - 947  2002

  • Learning Using Surface Fitting

    R. Shikishima, K. Hirasawa, J. Hu, M. Hashimoto

    SICE 2002   Osaka   1004 - 1009  2002

  • Studying the Effects of Multiplication Neurons for Parity Problems

    D. Li, K. Hirasawa, J. Hu, J. Murata

    SICE 2002   Osaka   2888 - 2891  2002

  • (N,ε) Stability Analysis of Nonlinear Systems Using Universal Learning Networks

    K. Hirasawa, J. Hu, J. Murata

    IFAC 2002   Barcelona  2002

  • A Quasi-ARX Model Incorporating Neural Network for Control of Nonlinear Systems

    J. Hu, K. Hirasawa, K. Kumamaru

    IFAC 2002   Barcelona  2002

  • Multiplication Units in Feed-Forward Neural Networks and Its Training

    D. Li, K. Hirasawa, J. Hu, J. Murata

    ICONIP 2002   Singapore  2002

    DOI

  • Multi-Branch Structure of Layered Neural Networks

    T. Yamasita, K. Hirasawa, J. Hu, J. Murata

    ICONIP 2002   Singapore  2002

    DOI

  • Auto-Associative Memory by Universal Learning Networks

    k.shibuta, k. Hirasawa, J. Hu, J.Murata

    ICONIP 2002   Singapore  2002

    DOI

  • A method for applying multilayer perceptrons to control of nonlinear systems

    Jinglu Hu, K. Hirasawa

    ICONIP 2002 - Proceedings of the 9th International Conference on Neural Information Processing: Computational Intelligence for the E-Age   3   1267 - 1271  2002

     View Summary

    This paper introduces a new method for applying multilayer perceptron (MLP) network to control of nonlinear systems. The MLP network is not used directly as a nonlinear controller, but used indirectly via an ARX-like macro-model. The ARX-like model incorporating MLP network is constructed in such a way that it has similar linear properties to a linear ARX model. The nonlinear controller is then designed in the same way as designing a linear controller based on a linear ARX model. Numerical simulations are carried to demonstrate the effectiveness of the new method.

    DOI

  • Automatic Generation of Boolean Functions Using Genetic Network Programming

    Y. Matsuya, K. Hirasawa, J. Hu, J. Murata

    SEAL 2002   Singapore  2002

  • 一般化学習ネットワークのインパルス応答に基ずく非線形制御方式

    平澤宏太郎, 橋本雅之, 胡敬炉, 村田純一, 金春樹

    電気学会論文誌C   122 ( 1 ) 105 - 115  2002

  • ニューラルネットワークを用いた非線形GPC

    浴百合雄, 村田純一, 平澤宏太郎

    電気学会論文誌C   122 ( 2 ) 265 - 269  2002

  • 遺伝的ネットワークプログラミングのオンライン学習

    間普真吾, 平澤宏太郎, 胡敬炉, 村田純一

    電気学会論文集C   122 ( 3 ) 355 - 362  2002

  • マルチエージェントシステムの共生学習進化(masbiole)の基礎検討

    平澤宏太郎, 吉田英正, 中西賢精, 胡敬炉

    電気学会論文誌C   122 ( 3 ) 346 - 354  2002

  • Enhancing the Generalization Ability of Backpropagation Algorithm through Controlling the Outputs of the Hidden Layers

    WAN Weishui, HIRASAWA Kotaro, HU Jinglu, MURATA Junichi

      38 ( 4 ) 411 - 419  2002

    DOI CiNii

  • Network Structure Oriented Evolutionary Model:

    H. Katagiri, K. Hirasawa, J. Hu, J. Murata, M. Kosaka

    計測自動制御学会論文集   38 ( 5 ) 485 - 494  2002

    DOI

  • Enhancement of RasID and Its Evaluation

    HIRASAWA Kotaro, MIYAZAKI Hiroyuki, HU Jinglu

    Transactions of the Society of Instrument and Control Engineers   38 ( 9 ) 775 - 783  2002

     View Summary

    In this paper, RasID is enhanced and evaluated systematically. RasID is an abbreviation of Random Search with Intensification and Diversification. RasID can search for a global minimum based on a probabilitiy density function, which can be modified adaptively using information based on success and failure of the past searching. As a result, RasID performs intensified search and diversified search iteratively to find a global minimum.<br>The improvements of RasID are, (1) the probability density function is modified so that each variable can have its own unique density function, and (2) multiple candidates for solutions are created in the search to enhance the efficiency of the searching.<br>Modified RasID is compared systematically with typical optimization methods such as Evolutionary. Programming and Fast Evolutionary Programming introduced recently using 23 different complibated functions. From simulation results, it has been clarified that the performance of Modified RasID is comparable to the ones of EP and FEP in spite of using one individual.

    DOI CiNii

  • Universal Learning Networks with Branch Control

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata, D. LI

    電気学会論文誌C   122 ( 10 ) 1812 - 1820  2002

  • 入力ゲート付ニューラルネットワークとそのエージェントの行動学習への影響

    村田純一, 鈴木政史, 平澤宏太郎

    電気学会論文集C   122 ( 11 ) 1969 - 1975  2002

  • Genetic Network Programming とそのマルチエージェントシステムへの応用

    片桐広信, 平澤宏太郎, 胡敬炉, 村田純一

    電気学会論文誌C   122 ( 12 ) 2149 - 2156  2002

  • An Extended Lotka-Volterra Model Based on Fuzzy Symbiosis and Fuzzy Spatial Distribution

    K. Hirasawa, J. Hu, N. Kusumi, S. Mabu

    J. Machine Intelligence and Robotic Control   4 ( 2 ) 49 - 59  2002

  • Robust Neural Controller Designing Method with a Dual Learning Algorithm

    H. Chen, K. Hirasawa, J. Hu

    J. Machine Intelligence and Robotic Control   4 ( 4 ) 135 - 142  2002

  • Online Learning of Genetic Network Programming(GNP)

    S. Mabu, K. Hirasawea, J. HU, J. Murata

    IEEE CEC 2002   Honolulu  2002

    DOI

  • Networks with Input Gates for Situation-Dependent Input Selection in Reinforecement Learning

    J. Murata, M.Suzuki, K. Hirasawa

    IJCNN 2002   Honolulu  2002

  • Learning of Symbiotic Relations among Agents by Using Neural Networks

    K. Hirasawa, H. Yoshida, K. Nakanishi

    IJCNN 2002   Honolulu  2002

  • The Dynamic Performance of Photovoltaic Supplied DC Motor Fed from DC-DC Converter by Neural Networks

    A. Hussein, K. Hirasawa, J. HU, J. Murata

    IJCNN 2002   Honolulu  2002

  • Training a Kind of Hybrid Universal Learning Networks with Classification Problems

    D. Li, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2002   Honolulu  2002

  • Adaptive Neural Network Speed Controller for DC Motor Drives

    A. Hussein, K. Hirasawa, J. Hu, J. Murata

    AdCONIP 2002   Kumamoto   465 - 470  2002

  • A Robust Controller Design Method of Nonlinear Systems for System Parameters Changes

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    AdCONIP 2002   Kumamoto   533 - 538  2002

  • A New Model to Realize Variable Size Genetic Network Programming-A Case Study with the Tileworld Problem

    H. Katagiri, K. Hirasawa, J. Hu, J. Murata

    GECCO 2002   New York   279 - 286  2002

  • Multiagent Systems with Symbiotic Learning and Evolution using Genetic Network Programming

    T. Eguchi, K. Hirasawa, J. Hu, J. Murata

    GECCO 2002   New York   130 - 137  2002

  • The Basic Study of Artificial Ecosystem Models Using Network-Type Assembly-Like Language

    Y. Shiraishi, K. Hirasawa, J. Hu, J. Murata

    GECCO 2002   New York   412 - 418  2002

  • Increasing Robustness of Genetic Algorithm

    J. Mao, K. Hoirasawa, J. Hu, J. Murata

    GECCO 2002   New York  2002

  • Auto Correlation Associative Memory by Universal Learning Networks

    K. Shibuta, K. Hirasawa, J. Hu, J. Murata

    SICE 2002   Osaka   793 - 798  2002

  • Multi-branch Structure of Layered Neural Networks

    T. Yamasita, K. Hirasawa, J. Hu, J. Murata

    SICE 2002   Osaka   799 - 804  2002

  • Task-Oriented Reiforcement Learning for Continuous Tasks in Dynamic Environment

    M. Kamal, J. Murata, K. Hirasawa

    SICE 2002   Osaka   932 - 935  2002

  • Comparing Some Graph Crossover in Genetic Network Programming

    H. Katagiri, K. Hirasawa, J. Hu, J. Murata

    SICE 2002   Osaka   936 - 941  2002

  • Automatic Generation of Programs Using Genetic Network Programming

    Y. Matsuya, K. Hirasawa, J. Hu, J. Murata

    SICE 2002   Osaka   942 - 947  2002

  • Learning Using Surface Fitting

    R. Shikishima, K. Hirasawa, J. Hu, M. Hashimoto

    SICE 2002   Osaka   1004 - 1009  2002

  • Studying the Effects of Multiplication Neurons for Parity Problems

    D. Li, K. Hirasawa, J. Hu, J. Murata

    SICE 2002   Osaka   2888 - 2891  2002

  • (N,ε) Stability Analysis of Nonlinear Systems Using Universal Learning Networks

    K. Hirasawa, J. Hu, J. Murata

    IFAC 2002   Barcelona  2002

  • A Quasi-ARX Model Incorporating Neural Network for Control of Nonlinear Systems

    J. Hu, K. Hirasawa, K. Kumamaru

    IFAC 2002   Barcelona  2002

  • Multiplication Units in Feed-Forward Neural Networks and Its Training

    D. Li, K. Hirasawa, J. Hu, J. Murata

    ICONIP 2002   Singapore  2002

    DOI

  • Multi-branch structure of layered neural networks

    T. Yamashita, K. Hirasawa, Jinglu Hu, J. Murata

    ICONIP 2002 - Proceedings of the 9th International Conference on Neural Information Processing: Computational Intelligence for the E-Age   1   243 - 247  2002

     View Summary

    In this paper, a multi-branch structure of neural networks is studied to make their size compact. The multi-branch structure has shown improved performance against conventional neural networks. As a result, it has been proved that the number of nodes of networks and the computational cost for training networks can be reduced. In addition, it could be said that proposed multi-branch networks are special cases of higher order neural networks, however, they obtain higher order effect easier without suffering the parameter explosion problem.

    DOI

  • Auto-Associative Memory by Universal Learning Networks

    k.shibuta, k. Hirasawa, J. Hu, J.Murata

    ICONIP 2002   Singapore  2002

    DOI

  • A method for Applying Multilayer Perceptrons to Control of Nonlinear Systems

    J. Hu, K. Hirasawa

    ICONIP 2002   Singapore  2002

    DOI

  • Automatic Generation of Boolean Functions Using Genetic Network Programming

    Y. Matsuya, K. Hirasawa, J. Hu, J. Murata

    SEAL 2002   Singapore  2002

  • Enhancing the Generalization Ability of Backpropagation Algorithm through Controlling the Outputs of Hidden Layers

    W. Wan, K. Hirasawa, J. Hu, J. Murata

    計測自動制御学会論文集   38 ( 4 ) 411 - 419  2002

    DOI CiNii

  • Network Structure Oriented Evolutionary Model:

    H. Katagiri, K. Hirasawa, J. Hu, J. Murata, M. Kosaka

    計測自動制御学会論文集   38 ( 5 ) 485 - 494  2002

    DOI

  • Universal Learning Networks with Branch Control

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata, D. LI

    電気学会論文誌C   122 ( 10 ) 1812 - 1820  2002

  • An Extended Lotka-Volterra Model Based on Fuzzy Symbiosis and Fuzzy Spatial Distribution

    K. Hirasawa, J. Hu, N. Kusumi, S. Mabu

    J. Machine Intelligence and Robotic Control   4 ( 2 ) 49 - 59  2002

  • Robust Neural Controller Designing Method with a Dual Learning Algorithm

    H. Chen, K. Hirasawa, J. Hu

    J. Machine Intelligence and Robotic Control   4 ( 4 ) 135 - 142  2002

  • Improvement of generalization ability for identifying dynamical systems by using universal learning networks

    K Hirasawa, S Kim, JL Hu, J Murata, M Han, CZ Jin

    NEURAL NETWORKS   14 ( 10 ) 1389 - 1404  2001.12

     View Summary

    This paper studies how the generalization ability of models of dynamical systems can be improved by taking advantage of the second order derivatives of the outputs with respect to the external inputs. The proposed method can be regarded as a direct implementation of the well-known regularization technique using the higher order derivatives of the Universal Learning Networks (ULN's). ULNs consist of a number of interconnected nodes where the nodes may have any continuously differentiable nonlinear functions in them and each pair of nodes can be connected by multiple branches with arbitrary time delays. A generalized learning algorithm has been derived for the ULNs, in which both the first order derivatives (gradients) and the higher order derivatives are incorporated. First, the method for computing the second order derivatives of ULNs is discussed. Then, a new method for implementing the regularization term is presented. Finally, simulation studies on identification of a nonlinear dynamical system with noises are carried out to demonstrate the effectiveness of the proposed method. Simulation results show that the proposed method can improve the generalization ability of neural networks significantly, especially in terms that (1) the robust network can be obtained even when the branches of trained ULNs are destructed, and (2) the obtained performance does not depend on the initial parameter values. (C) 2001 Elsevier Science Ltd. All rights reserved.

    DOI PubMed CiNii

  • Improvement of generalization ability for identifying dynamical systems by using universal learning networks

    K Hirasawa, S Kim, JL Hu, J Murata, M Han, CZ Jin

    NEURAL NETWORKS   14 ( 10 ) 1389 - 1404  2001.12

     View Summary

    This paper studies how the generalization ability of models of dynamical systems can be improved by taking advantage of the second order derivatives of the outputs with respect to the external inputs. The proposed method can be regarded as a direct implementation of the well-known regularization technique using the higher order derivatives of the Universal Learning Networks (ULN's). ULNs consist of a number of interconnected nodes where the nodes may have any continuously differentiable nonlinear functions in them and each pair of nodes can be connected by multiple branches with arbitrary time delays. A generalized learning algorithm has been derived for the ULNs, in which both the first order derivatives (gradients) and the higher order derivatives are incorporated. First, the method for computing the second order derivatives of ULNs is discussed. Then, a new method for implementing the regularization term is presented. Finally, simulation studies on identification of a nonlinear dynamical system with noises are carried out to demonstrate the effectiveness of the proposed method. Simulation results show that the proposed method can improve the generalization ability of neural networks significantly, especially in terms that (1) the robust network can be obtained even when the branches of trained ULNs are destructed, and (2) the obtained performance does not depend on the initial parameter values. (C) 2001 Elsevier Science Ltd. All rights reserved.

    DOI PubMed CiNii

  • A homotopy approach to improving PEM identification of ARMAX models

    JL Hu, K Hirasawa, K Kumamaru

    AUTOMATICA   37 ( 9 ) 1323 - 1334  2001.09

     View Summary

    This paper presents a homotopy approach to improving PEM identification of ARMAX model. PEM estimates of ARMAX model parameters are determined as the global minimum of criterion function, which is however not always unimodal because of the MA noise model part. An optimization-based PEM identification algorithm has a potential risk to be stuck at a local minimum that results in a poorly identified model. A homotopy continuation method is introduced to solve this problem. The idea is to start the estimation with the criterion function for PEM identification of the ARX model, which is gradually deformed into the actual one for PEM identification of the ARMAX model as the algorithm iterates. By building the deformation into the usual recursive procedure for the ARMAX identification and introducing a scheme to control the solution continuously staying in the global minima of the deformed criterion functions, the homotopy-based PEM identification algorithm is implemented in such a way that it has very good convergence performance, with only little increase in computation load compared to the usual PEM algorithm. (C) 2001 Elsevier Science Ltd. All rights reserved.

    DOI CiNii

  • A homotopy approach to improving PEM identification of ARMAX models

    Jinglu Hu, Kotaro Hirasawa, Kousuke Kumamaru

    Automatica   37 ( 9 ) 1323 - 1334  2001.09

     View Summary

    This paper presents a homotopy approach to improving PEM identification of ARMAX model. PEM estimates of ARMAX model parameters are determined as the global minimum of criterion function, which is however not always unimodal because of the MA noise model part. An optimization-based PEM identification algorithm has a potential risk to be stuck at a local minimum that results in a poorly identified model. A homotopy continuation method is introduced to solve this problem. The idea is to start the estimation with the criterion function for PEM identification of the ARX model, which is gradually deformed into the actual one for PEM identification of the ARMAX model as the algorithm iterates. By building the deformation into the usual recursive procedure for the ARMAX identification and introducing a scheme to control the solution continuously staying in the global minima of the deformed criterion functions, the homotopy-based PEM identification algorithm is implemented in such a way that it has very good convergence performance, with only little increase in computation load compared to the usual PEM algorithm. © 2001 Elsevier Science Ltd. All rights reserved.

    DOI CiNii

  • A new control method of nonlinear systems based on impulse responses of universal learning networks

    K Hirasawa, JL Hu, J Murata, CZ Jin

    IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART B-CYBERNETICS   31 ( 3 ) 362 - 372  2001.06

     View Summary

    A new control method of nonlinear dynamic systems is proposed based on the impulse responses of universal learning networks (ULNs). ULNs form a superset of neural networks. They consist of a number of interconnected nodes where the nodes may have any continuously differentiable nonlinear functions in them and each pair of nodes can be connected by multiple branches with arbitrary time delays. A generalized learning algorithm is derived for the ULNs, in which both the first order derivatives (gradients) and the higher order derivatives are incorporated. One of the distinguished features of the proposed control method is that the impulse response of the systems is considered as an extended part of the criterion function and it can be calculated by using the higher order derivatives of ULNs. By using the impulse response as the criterion function, nonlinear dynamics with not only quick response but also quick damping and small steady state error can be more easily obtained than the conventional nonlinear control systems with quadratic form criterion functions of state and control variables.

    DOI CiNii

  • A new control method of nonlinear systems based on impulse responses of universal learning networks

    Kotaro Hirasawa, Jinglu Hu, Junichi Murata, Chunzhi Jin

    IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics   31 ( 3 ) 362 - 372  2001.06

     View Summary

    A new control method of nonlinear dynamic systems is proposed based on the impulse responses of universal learning networks (ULNs). ULNs form a superset of neural networks. They consist of a number of interconnected nodes where the nodes may have any continuously differentiable nonlinear functions in them and each pair of nodes can be connected by multiple branches with arbitrary time delays. A generalized learning algorithm is derived for the ULNs, in which both the first order derivatives (gradients) and the higher order derivatives are incorporated. One of the distinguished features of the proposed control method is that the impulse response of the systems is considered as an extended part of the criterion function and it can be calculated by using the higher order derivatives of ULNs. By using the impulse response as the criterion function, nonlinear dynamics with not only quick response but also quick damping and small steady state error can be more easily obtained than the conventional nonlinear control systems with quadratic form criterion functions of state and control variables.

    DOI CiNii

  • GNPを用いた共生学習進化型マルチエージェントシステム

    江口徹, 平澤宏太郎, 胡敬炉, 村田純一

    SICE学術講演会   名古屋  2001

  • 共生学習進化型マルチエージェントシステム

    中西賢精, 平澤宏太郎, 胡敬炉, 村田純一

    SICE学術講演会   名古屋  2001

  • 一般化学習ネットワークによる連想記憶とその安定化

    渋田敬子, 平澤宏太郎, 胡敬炉, 村田純一

    SICE学術講演会   名古屋  2001

  • 評価値に応じた問題分割機能を持つゲート付きニューラルネットワーク

    梶原義龍, 村田純一, 平澤宏太郎

    SICE学術講演会   名古屋  2001

  • 遺伝的ネットワークプログラミングのオンライン学習

    間普真吾, 平澤宏太郎, 胡敬炉, 村田純一

    SICE学術講演会   名古屋  2001

  • ニューラルネットワークモデルの階層的学習

    胡敬炉, 平澤宏太郎

    SICE学術講演会   名古屋  2001

  • Stability Analysis of Nonlinear Systems Described by Universal Learning Networks

    K. Hirasawa, Y. Yu, J. Hu, J. Murata

    SICE学術講演会   名古屋  2001

  • A Study on a Modular Neural Network with Overlapping

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata

    FANシンポジウム   堺   449 - 454  2001

  • 評価値に応じた分割統治を行うゲート付きニューラルネットワーク

    梶原義龍, 村田純一, 平澤宏太郎

    FANシンポジウム   堺   445 - 448  2001

  • Universal Learning Networks with Multi-branches

    S. Kim, K. Hirasawa, J. Hu

    FANシンポジウム   堺   441 - 444  2001

  • 共生学習進化型マルチエージェントシステムにおける共生パレート解

    中西賢精, 平澤宏太郎, 胡敬炉, 村田純一

    FANシンポジウム   堺   225 - 228  2001

  • 共進化戦略を考慮したGenetic Network Programming

    大久保雅文, 平澤宏太郎, 胡敬炉

    FANシンポジウム   堺   41 - 44  2001

  • Genetic Network Programming and Its Application

    K. Hirasawa, H. Katagiri, J. Hu, J. Murata

    FANシンポジウム   堺   37 - 40  2001

  • 一般化学習ネットワークによる連想記憶モデルの記憶容量についての検討

    渋田敬子, 平澤宏太郎, 胡敬炉, 村田純一

    SICEシステム情報部門学術講演会   宮崎   459 - 461  2001

  • ダイナミカルノードゲート付きニューラルネットワーク

    馬場洋平, 村田純一, 平澤宏太郎

    SICEシステム情報部門学術講演会   宮崎   453 - 458  2001

  • Adaptive Neural Network Speed Controller for Photovoltaic Supplied

    A. Hussein, J. Hu, K. Hirasawa, J. Murata

    SICEシステム情報部門学術講演会   宮崎   425 - 430  2001

  • Genetic Network Programmingを用いた人工生態系モデルの基礎検討

    白石優旗, 平澤宏太郎, 胡敬炉, 村田純一

    SICEシステム情報部門学術講演会   宮崎   335 - 338  2001

  • ゲート付きニューラルネットワーク

    村田純一, 平澤宏太郎

    SICEシステム情報部門学術講演会   宮崎   301 - 306  2001

  • 線形特性を持つシグモイドニューラルネットワークの構成と応用

    胡敬炉, 平澤宏太郎

    SICEシステム情報部門学術講演会   宮崎   283 - 288  2001

  • Probabilistic Universal Learning Networks

    K. Hirasawa, J. Hu, J. Murata

    SICEシステム情報部門学術講演会   宮崎   265 - 270  2001

  • Relation between Weight Initialization of Neural Networks and Higher Order Learning Algorithm

    W. Wan, K. Hirasawa, J. Hu, J. Murasta

    SICEシステム情報部門学術講演会   宮崎   441 - 446  2001

  • 共生学習進化型マルチエージェントシステムにおけるパレート均衡解の探索

    江口徹, 平澤宏太郎, 胡敬炉, 村田純一

    SICEシステム情報部門学術講演会   宮崎   85 - 90  2001

  • Genetic Algorithm Based on Symbiotic Concept for Multiobjective Optimization Problem

    J. Mao, K. Hirasawa, J. Hu, J. Murata

    SICEシステム情報部門学術講演会   宮崎   49 - 54  2001

  • 遺伝的ネットワークプログラミングのオンライン強化学習

    間普真吾, 平澤宏太郎, 胡敬炉, 村田純一

    SICEシステム部門学術講演会   宮崎   25 - 30  2001

  • 強化学習エージェントの環境変化適応のための高次の知識と擬似報酬による学習高速化

    田村悠吉, 村田純一, 平澤宏太郎

    SICEシステム情報部門学術講演会   宮崎   19 - 24  2001

  • マーキングを用いた強化学習マルチエージェントシステムの性能向上

    藤木智博, 村田純一, 平澤宏太郎

    SICEシステム情報部門学術講演会   宮崎   1 - 6  2001

  • Task-Oriented Reiforcement Learning in Cooperative Multiagent System

    M. Kamal, J. Murata, K. Hirasawa

    SICE九州支部学術講演会   宮崎   477 - 480  2001

  • 強化学習マルチエージェントシステムにおける協力行動実現

    村田純一, 藤本雅貴, 平澤宏太郎

    SICE九州支部学術講演会   宮崎   453 - 456  2001

  • Comparison between Genetic Network Programming and Genetic Programming

    K. Hirasawa, M. Okubo, J. Hu, J. Murata

    SICE九州支部学術講演会   宮崎   427 - 430  2001

  • Study of Universal Learning Network with Branch Control

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata

    AROB 2000   Oita   470 - 473  2001

  • A New Method for Designing Robust Neural Network controller

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    AROB 2000   Oita   504 - 507  2001

  • Function of General Regulalization Terms-Case Study on Two-Spiral Classification Problem

    W. Wan, K. Hirasawa, J. Murata, J. Hu

    AROB 2000   Oita   516 - 519  2001

  • Function Approximation Using LVQ and Fuzzy Sets

    S. Min-Kyu, J. Murata, K. Hirasawa

    AROB 2000   Oita   520 - 523  2001

  • Comparison between Genetic Network Programming and Genetic Programming

    K. Hirasawa, M. Okubo, J. Hu, J. Murata

    CEC 2001   Seoul   1276 - 1282  2001

  • A Study of Functions Distribution of Neural Networks

    Q.Xiong, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2001   Washington DC   2361 - 2366  2001

  • Relation between Weight Initialization of Neural Networks and Pruning Algorithm

    W. Wan, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2001   Washington DC   1750 - 1755  2001

  • Enhancing the Generalization Ability of Neural Netyworks by Using Gram-Schmidt Orthogonalization Algorithm

    W. Wan, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2001   Washington DC   1721 - 1726  2001

  • An Embedded Neural Network For Modeling Nonlinear Systems

    J. Hu, K. Hirasawa

    IJCNN 2001   Washington DC   1698 - 1703  2001

  • Stability Analysis of Nonlinear Systems Using Higher Order Derivatives of Universal Learning Networks

    K. Hirasawa, Y. Yu, J. Hu, J. Murata

    IJCNN 2001   Washington DC   1273 - 1278  2001

  • Improvement of Generalization Ability for Identifying Dynamic Systems by Using Universal Learning Networks

    S. Kim, K. Hirasawa, J. Hu

    IJCNN 2001   Washington DC   1203 - 1208  2001

  • A New Robust Neural Network Controller Designing Method for Nonlinear Systems

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2001   Washington DC   497 - 502  2001

  • Universal Learning Networks with Multiplication Neurons and Its Representatyion Ability

    D. Li, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2001   Washington DC   150 - 155  2001

  • Genetic Symbiosis Algorithm for Multiobjective Optimization Problems

    J. Mao, K. Hirasawa, J. Hu, J. Murata

    GECCO 2001   San Fran cisco   267 - 274  2001

  • Network structure oriented evolutionary model-genetic network programming and its comparison with genetic programming

    KATAGIRI H.

    proc. of the 2001 Genetic and Evolutionary Computation Conference Late Breaking Papers, California, USA   San Fran cisco   219 - 226  2001

    CiNii

  • A hierarchical method for training embedded sigmoidal neural networks

    Jinglu Hu, Kotaro Hirasawa

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)   2130   937 - 942  2001

     View Summary

    This paper discusses the problem of applying sigmoidal neural networks to identification of nonlinear dynamical systems. When using sigmoidal neural networks directly as nonlinear models, one often meets problems such as model parameters lack of physical meaning, sensitivity to noise in model training. In this paper, we introduce an embedded sigmoidal neural network model, in which the neural network is not used directly as a model, but is embedded in a shield such that part of the model parameters become meaningful. Corresponding to the meaningful part and the meaningless part of model parameters, a hierarchical training algorithm consisting of two learning loops is then introduced to train the model. Simulation results show that such a dual loop learning algorithm can solve the noise sensitivity and local minimum problems to some extent.

    DOI

  • A study of Brain-like Neural Network Model

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata

    KES 2001   Osaka   303 - 307  2001

  • Function Approximation Using LVQ and Fuzzy Sets

    S. Min-Kyu, J. Murata, K. Hirasawa

    KES 2001   Osaka   829 - 833  2001

  • New Development on Tracking Algorithm with Derivative Measurement

    Dai, K. Hirasawa

    IEEE SMC 2001   Phoenix  2001

  • Function Approximation Using LVQ and Fuddy Sets

    S. Minkyu, J. Murata, K. Hirasawa

    IEEE SMC 2001   Phoenix   1442 - 1447  2001

  • Comparative Study between Functions Distributed Network and Ordinary Neural Network

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata

    IEEE SMC 2001   Phoenix   1548 - 1553  2001

  • 確率的ニューラルネットワークにおける自己組織化

    白石優旗, 平澤宏太郎, 胡敬炉, 村田純一

    電気学会論文集   121 ( 1 ) 187 - 195  2001

  • Learning of Symbiotic Phenomena between Agents by Using Nerural Networks

    HIRASAWA Kotaro, YOSHIDA Hidemasa, HU Jinglu

    The Transactions of the Institute of Electrical Engineers of Japan. C   121 ( 1 ) 177 - 186  2001

    CiNii

  • Universal Learning Networks with Varying Parameters Considering Branch Control

    HIRASAWA Kotaro, ETO Hironobu, HU Jinglu, MURATA Junichi, XIONG Quingyu

    The Transactions of the Institute of Electrical Engineers of Japan. C   121 ( 1 ) 98 - 105  2001

    CiNii

  • 蟻の行動進化におけるGenetic Network ProgrammingとGenetic Programmingの性能比較

    平澤宏太郎, 大久保雅文, 片桐広信, 胡敬炉, 村田純一

    電気学会論文集   121 ( 6 ) 1001 - 1009  2001

  • Modeling of Mutual Interaction of Complicated Systems by Spatial Distribution Universal Learning Networks

    KUSUMI Naohiro, HIRASAWA Kotaro, HU Jinglu, MURATA Junichi

    Transactions of the Society of Instrument and Control Engineers   37 ( 7 ) 657 - 664  2001

     View Summary

    Recently, many people are involved in the research of complex systems which are made of a large number of elements interacting with each other.<br>In this paper, a new modeling method of mutual interaction of the complicated systems is proposed by using Spatial Distribution Universal Learning Networks (SdULNs) and Fuzzy inference algorithms.<br>In addition, the proposed model is applied to the modeling of ecosystems which are described as the Lotka-Volterra equation.<br>From simulations where apriori information on ecosystems is given, it has been cleared that the proposed method can present more flexible emergent dynamics than the conventional Lotka-Volterra equation.

    DOI CiNii

  • A New Minimax Control Method for Nonlinear Systems Using Universal Learning Networks

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    電気学会論文集   121 ( 9 ) 1471 - 1478  2001

  • Genetic Symbiosis Algorithm for Multiobjective Optimization Problems

    MAO Jiangming, HIRASAWA Kotaro, HU Jinglu, MURATA Junichi

    Transactions of the Society of Instrument and Control Engineers   37 ( 9 ) 893 - 901  2001

     View Summary

    Evolutionary Algorithms are often well-suited for optimization problems. Since mid 1980's, the interest in multi-objective problems has been expanding rapidly. Various evolutionary algorithms for multiobjective problems have been developed which are capable of searching for multiple solutions concurrently in a single run. In this paper, we propose a new genetic symbiosis algorithm (GSA) for multiobjective optimization problems (MOP) based on the symbiotic concept found widely in ecosystems. In the proposed GSA for MOP, a set of symbiotic parameters are introduced to modify the fitness of individuals used for reproduction so as to obtain a variety of Pareto solutions corresponding to user's demands. The symbiotic parameters are trained by minimizing a user defined criterion function. Several numerical simulations are carried out to demonstrate the effectiveness of the proposed GSA.

    DOI CiNii

  • Behavior Learning of Autonomous Robots by Modified Learning Vector Quantization

    Shon MIN KYU, MURATA Junichi, HIRASAWA Kotaro

    Transactions of the Society of Instrument and Control Engineers   37 ( 12 ) 1162 - 1168  2001

     View Summary

    This paper presents a method for searching for the optimal paths for autonomously moving agents in mazes by modified Learning Vector Quantization (LVQ) in a reinforcement learning framework. LVQ algorithm is faster than Q-learning algorithms because LVQ concentrates on the best behavior in available behaviors while Q-learning algorithms calculate values of all available behaviors and choose the best behavior among them. However, ordinary LVQ sometimes mis-learns in the reinforcement learning environment due to erroneous teacher signals. Here a new LVQ algorithm is proposed to overcome this problem, which finds the optimal path more efficiently.

    DOI CiNii

  • Overlapped Multi-Neural-Network and Its Training Algorithm

    J. Hu, K. Hirasawa, Q. Xiong

    電気学会論文誌C   122 ( 12 ) 1949 - 1956  2001

  • Genetic Symbiosis Algorithm

    K. Hirasawa, J. Hu, J. Murata, J. Mao

    J. Machine Intelligence and Robotic Control   3 ( 1 ) 27 - 34  2001

  • A Quasi-ARMAX Approach to Modelling of Non-Linear Systems

    J. Hu, K. Kumamaru, K. Hirasawa

    INT. J. Control   74 ( 18 ) 1754 - 1766  2001

    DOI CiNii

  • Stability Analysis of Nonlinear Systems Described by Universal Learning Networks

    K. Hirasawa, Y. Yu, J. Hu, J. Murata

    SICE学術講演会   名古屋  2001

  • A Study on a Modular Neural Network with Overlapping

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata

    FANシンポジウム   堺   449 - 454  2001

  • Universal Learning Networks with Multi-branches

    S. Kim, K. Hirasawa, J. Hu

    FANシンポジウム   堺   441 - 444  2001

  • Genetic Network Programming and Its Application

    K. Hirasawa, H. Katagiri, J. Hu, J. Murata

    FANシンポジウム   堺   37 - 40  2001

  • Adaptive Neural Network Speed Controller for Photovoltaic Supplied

    A. Hussein, J. Hu, K. Hirasawa, J. Murata

    SICEシステム情報部門学術講演会   宮崎   425 - 430  2001

  • Probabilistic Universal Learning Networks

    K. Hirasawa, J. Hu, J. Murata

    SICEシステム情報部門学術講演会   宮崎   265 - 270  2001

  • Relation between Weight Initialization of Neural Networks and Higher Order Learning Algorithm

    W. Wan, K. Hirasawa, J. Hu, J. Murasta

    SICEシステム情報部門学術講演会   宮崎   441 - 446  2001

  • Genetic Algorithm Based on Symbiotic Concept for Multiobjective Optimization Problem

    J. Mao, K. Hirasawa, J. Hu, J. Murata

    SICEシステム情報部門学術講演会   宮崎   49 - 54  2001

  • Task-Oriented Reiforcement Learning in Cooperative Multiagent System

    M. Kamal, J. Murata, K. Hirasawa

    SICE九州支部学術講演会   宮崎   477 - 480  2001

  • Comparison between Genetic Network Programming and Genetic Programming

    K. Hirasawa, M. Okubo, J. Hu, J. Murata

    SICE九州支部学術講演会   宮崎   427 - 430  2001

  • Study of Universal Learning Network with Branch Control

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata

    AROB 2000   Oita   470 - 473  2001

  • A New Method for Designing Robust Neural Network controller

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    AROB 2000   Oita   504 - 507  2001

  • Function of General Regulalization Terms-Case Study on Two-Spiral Classification Problem

    W. Wan, K. Hirasawa, J. Murata, J. Hu

    AROB 2000   Oita   516 - 519  2001

  • Function Approximation Using LVQ and Fuzzy Sets

    S. Min-Kyu, J. Murata, K. Hirasawa

    AROB 2000   Oita   520 - 523  2001

  • Comparison between Genetic Network Programming and Genetic Programming

    K. Hirasawa, M. Okubo, J. Hu, J. Murata

    CEC 2001   Seoul   1276 - 1282  2001

  • A Study of Functions Distribution of Neural Networks

    Q.Xiong, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2001   Washington DC   2361 - 2366  2001

  • Relation between Weight Initialization of Neural Networks and Pruning Algorithm

    W. Wan, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2001   Washington DC   1750 - 1755  2001

  • Enhancing the Generalization Ability of Neural Netyworks by Using Gram-Schmidt Orthogonalization Algorithm

    W. Wan, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2001   Washington DC   1721 - 1726  2001

  • An Embedded Neural Network For Modeling Nonlinear Systems

    J. Hu, K. Hirasawa

    IJCNN 2001   Washington DC   1698 - 1703  2001

  • Stability Analysis of Nonlinear Systems Using Higher Order Derivatives of Universal Learning Networks

    K. Hirasawa, Y. Yu, J. Hu, J. Murata

    IJCNN 2001   Washington DC   1273 - 1278  2001

  • Improvement of Generalization Ability for Identifying Dynamic Systems by Using Universal Learning Networks

    S. Kim, K. Hirasawa, J. Hu

    IJCNN 2001   Washington DC   1203 - 1208  2001

  • A New Robust Neural Network Controller Designing Method for Nonlinear Systems

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2001   Washington DC   497 - 502  2001

  • Universal Learning Networks with Multiplication Neurons and Its Representatyion Ability

    D. Li, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2001   Washington DC   150 - 155  2001

  • Genetic Symbiosis Algorithm for Multiobjective Optimization Problems

    J. Mao, K. Hirasawa, J. Hu, J. Murata

    GECCO 2001   San Fran cisco   267 - 274  2001

  • Network Structure Oriented Evolutionary Model-Genetic Network Programming and Its Comparison with Genetic Programming

    H. Katagiri, K. Hirasawa, J. Hu, J. Murata

    GECCO 2001   San Fran cisco   219 - 226  2001

  • A Hierarchical Method for Training Embedded Sigmoidal Neural Networks

    J. Hu, K. Hirasawa

    ICANN 2001   Viena  2001

    DOI

  • A study of Brain-like Neural Network Model

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata

    KES 2001   Osaka   303 - 307  2001

  • Function Approximation Using LVQ and Fuzzy Sets

    S. Min-Kyu, J. Murata, K. Hirasawa

    KES 2001   Osaka   829 - 833  2001

  • New Development on Tracking Algorithm with Derivative Measurement

    Dai, K. Hirasawa

    IEEE SMC 2001   Phoenix  2001

  • Function Approximation Using LVQ and Fuddy Sets

    S. Minkyu, J. Murata, K. Hirasawa

    IEEE SMC 2001   Phoenix   1442 - 1447  2001

  • Comparative Study between Functions Distributed Network and Ordinary Neural Network

    Q. Xiong, K. Hirasawa, J. Hu, J. Murata

    IEEE SMC 2001   Phoenix   1548 - 1553  2001

  • A New Minimax Control Method for Nonlinear Systems Using Universal Learning Networks

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    電気学会論文集   121 ( 9 ) 1471 - 1478  2001

  • Genetic Symbiosis Algorithm for Multiobjective Optimization Problems

    J. Mao, K. Hirasawa, J. Hu, J. Murata

    計測自動制御学会論文集   37 ( 9 ) 893 - 901  2001

     View Summary

    Evolutionary Algorithms are often well-suited for optimization problems. Since mid 1980's, the interest in multi-objective problems has been expanding rapidly. Various evolutionary algorithms for multiobjective problems have been developed which are capable of searching for multiple solutions concurrently in a single run. In this paper, we propose a new genetic symbiosis algorithm (GSA) for multiobjective optimization problems (MOP) based on the symbiotic concept found widely in ecosystems. In the proposed GSA for MOP, a set of symbiotic parameters are introduced to modify the fitness of individuals used for reproduction so as to obtain a variety of Pareto solutions corresponding to user's demands. The symbiotic parameters are trained by minimizing a user defined criterion function. Several numerical simulations are carried out to demonstrate the effectiveness of the proposed GSA.

    DOI CiNii

  • Behavior Learning of Autonomous Robots by Modified Learning Vector Quantization

    S. Min Kyu, J. Murata, K. Hirasawa

    計測自動制御学会論文集   37 ( 12 ) 1162 - 1168  2001

     View Summary

    This paper presents a method for searching for the optimal paths for autonomously moving agents in mazes by modified Learning Vector Quantization (LVQ) in a reinforcement learning framework. LVQ algorithm is faster than Q-learning algorithms because LVQ concentrates on the best behavior in available behaviors while Q-learning algorithms calculate values of all available behaviors and choose the best behavior among them. However, ordinary LVQ sometimes mis-learns in the reinforcement learning environment due to erroneous teacher signals. Here a new LVQ algorithm is proposed to overcome this problem, which finds the optimal path more efficiently.

    DOI CiNii

  • Overlapped Multi-Neural-Network and Its Training Algorithm

    J. Hu, K. Hirasawa, Q. Xiong

    電気学会論文誌C   122 ( 12 ) 1949 - 1956  2001

  • Genetic Symbiosis Algorithm

    K. Hirasawa, J. Hu, J. Murata, J. Mao

    J. Machine Intelligence and Robotic Control   3 ( 1 ) 27 - 34  2001

  • A Quasi-ARMAX Approach to Modelling of Non-Linear Systems

    J. Hu, K. Kumamaru, K. Hirasawa

    INT. J. Control   74 ( 18 ) 1754 - 1766  2001

    DOI CiNii

  • Universal learning network and its application to robust control

    K Hirasawa, J Murata, J Hu, C Jin

    IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART B-CYBERNETICS   30 ( 3 ) 419 - 430  2000.06

     View Summary

    Universal learning networks (ULN's) and robust control system design are discussed, ULN's provide a generalized framework to model and control complex systems. They consist of a number of inter-connected nodes where the nodes may have any continuously differentiable nonlinear functions in them and each pair of nodes can be connected by multiple branches with arbitrary time delays. Therefore, physical systems which can be described by differential or difference equations and also their controllers can be modeled in a unified way. So, ULN's constitute a superset of neural networks or fuzzy neural networks. In order to optimize the systems, a generalized learning algorithm is derived for the ULN's, in which both the first order derivatives (gradients) and the higher order derivatives are incorporated. The derivatives are calculated by using forward or backward propagation schemes. These algorithms for calculating the derivatives are extended versions of back propagation through time (BPTT) and real time recurrent learning (RTRL) by Williams in the sense that generalized nonlinear functions and higher order derivatives are dealt with. As an application of ULN's, the higher order derivative, one of the distinguished features of ULN's, is applied to realizing a robust control system in this paper, In addition, it is shown that the higher order derivatives are effective tools to realize sophisticated control of nonlinear systems. Other features of ULN's such as multiple branches with arbitrary time delays and using a priori information will be discussed in other papers.

    DOI CiNii

  • Universal learning network and its application to chaos control

    K Hirasawa, XF Wang, J Murata, JL Hu, CZ Jin

    NEURAL NETWORKS   13 ( 2 ) 239 - 253  2000.03

     View Summary

    Universal Learning Networks (ULNs) are proposed and their application to chaos control is discussed. ULNs provide a generalized framework to model and control complex systems. They consist of a number of inter-connected nodes where the nodes may have any continuously differentiable nonlinear functions in them and each pair of nodes can be connected by multiple branches with arbitrary time delays. Therefore, physical systems, which can be described by differential or difference equations and also their controllers, can be modeled in a unified way, and so ULNs may form a super set of neural networks and fuzzy neural networks. In order to optimize the ULNs, a generalized learning algorithm is derived, in which both the first order derivatives (gradients) and the higher order derivatives are incorporated. The derivatives are calculated by using forward or backward propagation schemes. These algorithms for calculating the derivatives are extended versions of Back Propagation Through Time (BPTT) and Real Time Recurrent Learning (RTRL) of Williams in the sense that generalized node functions, generalized network connections with multi-branch of arbitrary time delays, generalized criterion functions and higher order derivatives can be deal with. As an application of ULNs, a chaos control method using maximum Lyapunov exponent of ULNs is proposed. Maximum Lyapunov exponent of ULNs can be formulated by using higher order derivatives of ULNs, and the parameters of ULNs can be adjusted so that the maximum Lyapunov exponent approaches the target value. From the simulation results, it has been shown that a fully connected ULN with three nodes is able to display chaotic behaviors. (C) 2000 Elsevier Science Ltd. All rights reserved.

    DOI PubMed CiNii

  • Universal learning network and its application to chaos control

    K Hirasawa, XF Wang, J Murata, JL Hu, CZ Jin

    NEURAL NETWORKS   13 ( 2 ) 239 - 253  2000.03

     View Summary

    Universal Learning Networks (ULNs) are proposed and their application to chaos control is discussed. ULNs provide a generalized framework to model and control complex systems. They consist of a number of inter-connected nodes where the nodes may have any continuously differentiable nonlinear functions in them and each pair of nodes can be connected by multiple branches with arbitrary time delays. Therefore, physical systems, which can be described by differential or difference equations and also their controllers, can be modeled in a unified way, and so ULNs may form a super set of neural networks and fuzzy neural networks. In order to optimize the ULNs, a generalized learning algorithm is derived, in which both the first order derivatives (gradients) and the higher order derivatives are incorporated. The derivatives are calculated by using forward or backward propagation schemes. These algorithms for calculating the derivatives are extended versions of Back Propagation Through Time (BPTT) and Real Time Recurrent Learning (RTRL) of Williams in the sense that generalized node functions, generalized network connections with multi-branch of arbitrary time delays, generalized criterion functions and higher order derivatives can be deal with. As an application of ULNs, a chaos control method using maximum Lyapunov exponent of ULNs is proposed. Maximum Lyapunov exponent of ULNs can be formulated by using higher order derivatives of ULNs, and the parameters of ULNs can be adjusted so that the maximum Lyapunov exponent approaches the target value. From the simulation results, it has been shown that a fully connected ULN with three nodes is able to display chaotic behaviors. (C) 2000 Elsevier Science Ltd. All rights reserved.

    DOI PubMed CiNii

  • Genetic Network Programingとその応用

    片桐信広, 平澤宏太郎, 胡敬炉

    自律分散シンポジウム   沖縄   407 - 410  2000

  • 共生と進化現象のモデリング

    武居雅暁, 平澤宏太郎, 胡敬炉

    自律分散シンポジウム   沖縄   433 - 438  2000

  • 確率一般化学習ネットワークとその非線形制御システムへの応用

    平澤宏太郎, 金春樹, 村田純一, 胡敬炉

    情報処理学会研究会   飯塚   9 - 12  2000

  • Neural Networks with Node Gates

    H. Myint, J. Murata, T. Nakazono, K. Hirasawa

    システム制御情報学会研究講演発表会   京都   467 - 468  2000

  • 入力ゲート付きニューラルネットワークとそのエージェントの行動学習への応用

    鈴木政史, 村田純一, 平澤宏太郎

    SICE学術講演会   飯塚  2000

  • Genetic Network Programmingとその応用システム

    片桐広信, 平澤宏太郎, 胡敬炉, 村田純一

    SICE学術講演会   飯塚  2000

  • Neural Network with Node Gates and its Application to Nonlinear System Control

    MURATA Junichi, KAKIHARA Tomohide, FUJIMOTO Masaki

    Research reports on information science and electrical engineering of Kyushu University   飯塚 ( 2 ) 243 - 248  2000

    DOI CiNii

  • 確率的ニューラルネットワークにおける自己組織化

    白石優旗, 平澤宏太郎, 胡敬炉, 村田純一

    SICE学術講演会   飯塚  2000

  • 共生と進化現象を統合する生態系のモデリング手法の研究

    平山雄也, 平澤宏太郎, 胡敬炉

    SICE学術講演会   飯塚  2000

  • 共生進化を考慮したマルチエージェントシステム

    中西賢精, 平澤宏太郎, 胡敬炉

    SICE学術講演会   飯塚  2000

  • Genetic Network Programmingと蟻の行動の進化への応用

    大久保雅文, 平澤宏太郎, 胡敬炉

    SICE学術講演会   飯塚  2000

  • 複数の分散探索、集中探索エージェントを用いた最適化手法

    伊藤信治, 村田純一, 平澤宏太郎

    FANシンポジウム   東京  2000

  • ファジーとLVQを用いた関数近似

    孫敏圭, 村田純一, 平澤宏太郎

    FANシンポジウム   東京   313 - 316  2000

  • ニューラルネットワークによる共生進化マルチエージェントシステムの研究

    中西賢精, 平澤宏太郎, 胡敬炉, 村田純一

    FANシンポジウム   東京   235 - 238  2000

  • Genetic Network ProgrammingとGenetic Programming の性能比較

    大久保雅文, 平澤宏太郎, 胡敬炉

    FANシンポジウム   東京   433 - 436  2000

  • Genetic Network Programmingの性能評価

    大久保雅文, 平澤宏太郎, 胡敬炉

    SICE九州支部学術講演会   宮崎   471 - 474  2000

  • Improvement of Generalization Ability for Identifying Dynamic Systems by Using Universal Learning Networks

    S. Kim, K. Hirasawa, J. Hu

    SICE九州支部学術講演会   宮崎   467 - 470  2000

  • 共生進化型マルチエージェントシステムの性能評価

    中西賢精, 平澤宏太郎, 胡敬炉, 村田純一

    SICE九州支部学術講演会   宮崎  2000

  • 入力ゲート付きニューラルネットワークとそのエージェントの行動学習への応用

    鈴木政史, 村田純一, 平澤宏太郎

    SICEシステム情報シンポジウム   大阪   7 - 11  2000

  • 強化学習エージェントの環境変化適応のための高次の知識の抽出と利用

    田村悠吉, 村田純一, 平澤宏太郎

    SICEシステム情報シンポジウム   大阪   273 - 277  2000

  • 適応的離散ランダム探索法RasID-Dの組合わせ最適化問題への適用

    宮崎弘幸, 平澤宏太郎, 胡敬炉

    SICEシステム情報シンポジウム   大阪   327 - 332  2000

  • ニューラルネットワークによるエージェント間の共生現象の学習

    吉田英正, 平澤宏太郎, 胡敬炉

    SICEシステム情報シンポジウム   大阪   267 - 272  2000

  • Universal Learning Network and Its Application

    K. Hirasawa, J. Murata, J. Hu

    SICEシステム情報シンポジウム   大阪   127 - 132  2000

  • A Quasi-Modular Network: Neural Network with Function Localization

    J. Hu, K. Hirasawa

    SICEシステム情報シンポジウム   大阪   293 - 298  2000

  • A Method for Design of a Robust Controller of Nonlinear Systems

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    SICEシステム情報シンポジウム   大阪   289 - 292  2000

  • Genetic Symbiosis Algorithm for Multi-objective Optimization Problems

    J. Mao, K. Hirasawa, J. Hu, J. Murata

    SICEシステム情報シンポジウム   大阪   13 - 18  2000

  • Quasi-ARMAX Modeling Approaches to Identification and Prediction of Nonlinear Systems

    J. Hu, K. Kumamaru, K. Hirasawa

    IFAC Symp. on Identification 2000   Santa Barbara  2000

  • Stability analysis of inverted pendulum control by using Universal Learning Networks

    YU Y.

    Proc. of 2000 Asian Automatic Control Conference   Shanghai   2147 - 2152  2000

    CiNii

  • Synthesis of Functions Distributed Neural Networks Using Kohonen's Self-organizing Maps

    Q. Xiong, K. Hirasawa, J. Hu

    Asian Control Conference 2000   Shanghai   1999 - 2004  2000

  • Behavior Learning of Autonomous Robots by Modified Learning Vector Quantization

    S. Min-Kyu, J. Murata, K. Hirasawa

    Asian Control Conference 2000   Shanghai   1354 - 1359  2000

  • Genetic Symbiosis Algorithm

    K. Hirasawa, Y. Ishikawa, J. Hu, J. Murata, J. Mao

    IEEE CEC 2000   San Diego   1377 - 1384  2000

  • A New Method to Prune the Neural Network

    W. Wan, K. Hirasawa, J. Hu, C. Jin

    IJCNN 2000   Washington DC   449 - 454  2000

  • Universal Learning Networks with Branch Control

    K. Hirasawa, J. Hu, Q. Qiong, J. Murata, Y. Shiraishi

    IJCNN 2000   Washington DC   97 - 102  2000

  • Min Max Control of Nonlinear Systems Using Universal Learning Networks

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2000   Washington DC   242 - 247  2000

  • Overlapped Multi-Neural Network: A Case Study

    J. Hu, K. Hirasawa

    IJCNN 2000   Washington DC   120 - 125  2000

  • Stability Evaluation of Inverted Pendulum Control by Universal Learning Networks

    Y. Yu, K. Hirasawa, J. Hu, J. Murata

    Int. Conference Electrical Engineering   Kitakyushu   287 - 290  2000

  • Rasid Training of Multi Agent Systems with Fuzzy Inference-Based Mutual Interactions

    K. Hirasawa, J. Misawa, J. Hu, H. Katagiri, J. Murata

    SAB 2000   Paris   275 - 284  2000

  • Growing RBF structures using self-organizing maps

    QY Xiong, K Hirasawa, JL Hu, J Murata

    IEEE RO-MAN 2000: 9TH IEEE INTERNATIONAL WORKSHOP ON ROBOT AND HUMAN INTERACTIVE COMMUNICATION, PROCEEDINGS   Osaka   107 - 111  2000

     View Summary

    We present a novel growing RBF network structure using SOM in this paper. It consists of SOM and RBF networks respectively. The SOM performs unsupervised learning and also the weight vectors belonging to its out-put nodes are transmitted to the hidden nodes in the RBF networks as the centers of RBF activation functions, as a result one to one correspondence relationship is realized between the output nodes in SOM and the hidden nodes in RBF networks. The RBF networks perform supervised training using delta rule. Therefore, the current output errors in the RBF networks can be used to determine where to insert a new SOM unit according to the rule. This also makes it possible to make the RBF networks grow until a performance criterion is fulfilled or until a desired network size is obtained. The simulations on the two-spirals benchmark are shown to prove the proposed networks have good performance.

    DOI

  • Genetic symbiosis algorithm for multiobjective optimization problem

    JM Mao, K Hirasawa, JL Hu, J Murata

    IEEE RO-MAN 2000: 9TH IEEE INTERNATIONAL WORKSHOP ON ROBOT AND HUMAN INTERACTIVE COMMUNICATION, PROCEEDINGS   Osaka   137 - 142  2000

     View Summary

    Evolutionary Algorithms are often well-suited for optimization problems. Since the mid-1980's, interest in multiobjective problems has been expanding rapidly. Various evolutionary algorithms have been developed which are capable of searching for multiple solutions concurrently in a single run. In this paper, we proposed a genetic symbiosis algorithm (GSA) for multi-object optimization problems (MOP) based on the symbiotic concept found widely in ecosystem. In the proposed CSA for MOP, a set of symbiotic parameters are introduced to modify the fitness of individuals used for reproduction so as to obtain a variety of Pareto solutions corresponding to user's demands. The symbiotic parameters are trained by minimizing a user defined criterion function. Several numerical simulations are carried out to demonstrate the effectiveness of proposed GSA.

    DOI

  • Neural networks with node gates

    HM Myint, J Murata, T Nakazono, K Hirasawa

    IEEE RO-MAN 2000: 9TH IEEE INTERNATIONAL WORKSHOP ON ROBOT AND HUMAN INTERACTIVE COMMUNICATION, PROCEEDINGS   Osaka   253 - 257  2000

     View Summary

    Function approximation problems for the ordinary neural networks may be rather difficult if the function becomes complicated, due to the necessity of big network size and the possibilities of many local minima. A promissing way to solve these difficulties a's the localization of the problem. According to this concept, a new architecture of neural network is proposed namely neural network with node gates. In this paper, a function approximation example is provided to demonstrate the better performance of the proposed networks than the ordinary neural network.

    DOI

  • Determination of the Appropriate Node Function of NNs by Using Cascade-Correlation Algorithms

    W. Wan, K. Hirasawa, J. Murata

    IEEE IES 2000   Tokyo   1177 - 1182  2000

  • Optimal Structure Analysis of Universal Learning Networks with Multi-branch

    M. Han, K. Hirasawa, H. Ni, X. Jia

    IEEE SMC 2000   Nashvile   3171 - 3176  2000

  • Self-organization in Probabilistic Neural Networks

    Y. Shiraishi, K. Hirasawa, J. Hu, J. Murata

    IEEE SMC 2001   Nashvile   2533 - 2538  2000

  • Nonlinear Model Predictive Control Using a Neuro-Fuzzy Predictor

    J. Waller, J. Hu, K. Hirasawa

    IEEE SMC 2000   Nashvile   3459 - 3464  2000

  • Genetic Network Programming -Application to Intelligent Agents

    H. Katagiri, K. Hirasawa, J. Hu

    IEEE SMC 2000   Nashvile   3829 - 3834  2000

  • 発電用石炭だきボイラのための適応状態推定システム

    深山穂, 平澤宏太郎, 村上義雄, 津村俊一

    電気学会論文誌   120 ( 7 ) 993 - 1002  2000

  • A New Method Based on Determining Error Surface for Designing Three Layer Neural Networks

    LU Baiquan, MURATA Junichi, HIRASAWA Kotaro

    Transactions of the Society of Instrument and Control Engineers   36 ( 7 ) 589 - 598  2000

     View Summary

    A method is proposed for designing three layer neural networks that gives relevant network structures which assure global minimization of learning errors for small training sets and small learning errors for big training sets both irrespective of the initial values. A condition on network structure is considered to achieve the above purpose, and a number of possible structures are provided together with their learning algorithms. Also, the generalization abilities of the network structures are discussed to guide the choice of structures in practice. All of the proposed structures for small training sets have zero errors after learning by a gradient-based algorithm and thus solve the local minima problem. The difference between them is in the level of locality and generalization abilities. For a big training set, first, the structure with zero learning errors for part of training data is obtained, then all of training data are used to train the network of the given structure, which improves the generalization abilities. Numerical examples are provided that support the present approach.

    DOI CiNii

  • 確率一般化学習ネットワークによる非線形動的システムの同定

    平澤宏太郎, 四元一章, 胡敬炉, 干雲青

    電気学会論文集   120 ( 10 ) 1380 - 1387  2000

  • 確率一般化学習ネットワークとその非線形制御システムへの応用

    金春樹, 平澤宏太郎, 胡敬炉, 村田純一, 松岡拓哉

    情報処理学会論文誌:数理モデル化と応用   41 ( SIG7 ) 64 - 78  2000

  • RasID Training of Multi Agent Systems with Fuzzy Inference-Based Mutual Interactions

    K. Hirasawa, J. Hu, J. Misawa, J. Murata, H. Katagiri

    J. Machine Intelligence and Robotic Control   2 ( 1 ) 17 - 25  2000

  • Chaos control on universal learning networks

    Kotaro Hirasawa, Junichi Murata, Jinglu Hu, Chunzhi Jin

    IEEE Transactions on Systems, Man and Cybernetics Part C: Applications and Reviews   30 ( 1 ) 95 - 104  2000

     View Summary

    A new chaos control method is proposed which is useful for taking advantage of chaos and avoiding it. The proposed method is based on the following facts: 1) chaotic phenomena can be generated and eliminated by controlling maximum Lyapunov exponent of systems and 2) maximum Lyapunov exponent can be formulated and calculated by using higher order derivatives of Universal Learning Networks (ULN's). ULN's consist of a number of inter-connected nodes where the nodes may have any continuously differentiable nonlinear functions in them and each pair of nodes can be connected by multiple branches with arbitrary time delays. A generalized learning algorithm has been derived for the ULN's, in which both the first-order derivatives (gradients) and the higher order derivatives are incorporated. In simulations, parameters of ULN's with bounded node outputs are adjusted for maximum Lyapunov component to approach the target value. And, it has been shown that a fully connected ULN with three sigmoidal function nodes is able to generate and eliminate chaotic behaviors by adjusting the parameters.

    DOI CiNii

  • Stability Analysis of Robust Control Using Higher Order Derivatives of Universal Learning Networks

    Y. Yu, K. Hirasawa, J. Hu, J. Murata

    J. Machine Intelligence and Robotic Control   2 ( 3 ) 117 - 127  2000

    DOI

  • Neural Networks with Node Gates

    H. Myint, J. Murata, T. Nakazono, K. Hirasawa

    システム制御情報学会研究講演発表会   京都   467 - 468  2000

  • Improvement of Generalization Ability for Identifying Dynamic Systems by Using Universal Learning Networks

    S. Kim, K. Hirasawa, J. Hu

    SICE九州支部学術講演会   宮崎   467 - 470  2000

  • Universal Learning Network and Its Application

    K. Hirasawa, J. Murata, J. Hu

    SICEシステム情報シンポジウム   大阪   127 - 132  2000

  • A Quasi-Modular Network: Neural Network with Function Localization

    J. Hu, K. Hirasawa

    SICEシステム情報シンポジウム   大阪   293 - 298  2000

  • A Method for Design of a Robust Controller of Nonlinear Systems

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    SICEシステム情報シンポジウム   大阪   289 - 292  2000

  • Genetic Symbiosis Algorithm for Multi-objective Optimization Problems

    J. Mao, K. Hirasawa, J. Hu, J. Murata

    SICEシステム情報シンポジウム   大阪   13 - 18  2000

  • Quasi-ARMAX Modeling Approaches to Identification and Prediction of Nonlinear Systems

    J. Hu, K. Kumamaru, K. Hirasawa

    IFAC Symp. on Identification 2000   Santa Barbara  2000

  • Stability Analysis of Inverted Pendulum Control by Using Universal Learning Networks

    Y. Yu, K. Hirasawa, J. Hu, J. Murata

    Asian Control Conference 2000   Shanghai   2147 - 2152  2000

  • Synthesis of Functions Distributed Neural Networks Using Kohonen's Self-organizing Maps

    Q. Xiong, K. Hirasawa, J. Hu

    Asian Control Conference 2000   Shanghai   1999 - 2004  2000

  • Behavior Learning of Autonomous Robots by Modified Learning Vector Quantization

    S. Min-Kyu, J. Murata, K. Hirasawa

    Asian Control Conference 2000   Shanghai   1354 - 1359  2000

  • Genetic Symbiosis Algorithm

    K. Hirasawa, Y. Ishikawa, J. Hu, J. Murata, J. Mao

    IEEE CEC 2000   San Diego   1377 - 1384  2000

  • A New Method to Prune the Neural Network

    W. Wan, K. Hirasawa, J. Hu, C. Jin

    IJCNN 2000   Washington DC   449 - 454  2000

  • Universal Learning Networks with Branch Control

    K. Hirasawa, J. Hu, Q. Qiong, J. Murata, Y. Shiraishi

    IJCNN 2000   Washington DC   97 - 102  2000

  • Min Max Control of Nonlinear Systems Using Universal Learning Networks

    H. Chen, K. Hirasawa, J. Hu, J. Murata

    IJCNN 2000   Washington DC   242 - 247  2000

  • Overlapped Multi-Neural Network: A Case Study

    J. Hu, K. Hirasawa

    IJCNN 2000   Washington DC   120 - 125  2000

  • Stability Evaluation of Inverted Pendulum Control by Universal Learning Networks

    Y. Yu, K. Hirasawa, J. Hu, J. Murata

    Int. Conference Electrical Engineering   Kitakyushu   287 - 290  2000

  • Rasid Training of Multi Agent Systems with Fuzzy Inference-Based Mutual Interactions

    K. Hirasawa, J. Misawa, J. Hu, H. Katagiri, J. Murata

    SAB 2000   Paris   275 - 284  2000

  • Growing RBF structures using self-organizing maps

    QY Xiong, K Hirasawa, JL Hu, J Murata

    IEEE RO-MAN 2000: 9TH IEEE INTERNATIONAL WORKSHOP ON ROBOT AND HUMAN INTERACTIVE COMMUNICATION, PROCEEDINGS   Osaka   107 - 111  2000

     View Summary

    We present a novel growing RBF network structure using SOM in this paper. It consists of SOM and RBF networks respectively. The SOM performs unsupervised learning and also the weight vectors belonging to its out-put nodes are transmitted to the hidden nodes in the RBF networks as the centers of RBF activation functions, as a result one to one correspondence relationship is realized between the output nodes in SOM and the hidden nodes in RBF networks. The RBF networks perform supervised training using delta rule. Therefore, the current output errors in the RBF networks can be used to determine where to insert a new SOM unit according to the rule. This also makes it possible to make the RBF networks grow until a performance criterion is fulfilled or until a desired network size is obtained. The simulations on the two-spirals benchmark are shown to prove the proposed networks have good performance.

    DOI

  • Genetic symbiosis algorithm for multiobjective optimization problem

    JM Mao, K Hirasawa, JL Hu, J Murata

    IEEE RO-MAN 2000: 9TH IEEE INTERNATIONAL WORKSHOP ON ROBOT AND HUMAN INTERACTIVE COMMUNICATION, PROCEEDINGS   Osaka   137 - 142  2000

     View Summary

    Evolutionary Algorithms are often well-suited for optimization problems. Since the mid-1980's, interest in multiobjective problems has been expanding rapidly. Various evolutionary algorithms have been developed which are capable of searching for multiple solutions concurrently in a single run. In this paper, we proposed a genetic symbiosis algorithm (GSA) for multi-object optimization problems (MOP) based on the symbiotic concept found widely in ecosystem. In the proposed CSA for MOP, a set of symbiotic parameters are introduced to modify the fitness of individuals used for reproduction so as to obtain a variety of Pareto solutions corresponding to user's demands. The symbiotic parameters are trained by minimizing a user defined criterion function. Several numerical simulations are carried out to demonstrate the effectiveness of proposed GSA.

    DOI

  • Neural networks with node gates

    HM Myint, J Murata, T Nakazono, K Hirasawa

    IEEE RO-MAN 2000: 9TH IEEE INTERNATIONAL WORKSHOP ON ROBOT AND HUMAN INTERACTIVE COMMUNICATION, PROCEEDINGS   Osaka   253 - 257  2000

     View Summary

    Function approximation problems for the ordinary neural networks may be rather difficult if the function becomes complicated, due to the necessity of big network size and the possibilities of many local minima. A promissing way to solve these difficulties a's the localization of the problem. According to this concept, a new architecture of neural network is proposed namely neural network with node gates. In this paper, a function approximation example is provided to demonstrate the better performance of the proposed networks than the ordinary neural network.

    DOI

  • Determination of the Appropriate Node Function of NNs by Using Cascade-Correlation Algorithms

    W. Wan, K. Hirasawa, J. Murata

    IEEE IES 2000   Tokyo   1177 - 1182  2000

  • Optimal Structure Analysis of Universal Learning Networks with Multi-branch

    M. Han, K. Hirasawa, H. Ni, X. Jia

    IEEE SMC 2000   Nashvile   3171 - 3176  2000

  • Self-organization in Probabilistic Neural Networks

    Y. Shiraishi, K. Hirasawa, J. Hu, J. Murata

    IEEE SMC 2001   Nashvile   2533 - 2538  2000

  • Nonlinear Model Predictive Control Using a Neuro-Fuzzy Predictor

    J. Waller, J. Hu, K. Hirasawa

    IEEE SMC 2000   Nashvile   3459 - 3464  2000

  • Genetic Network Programming -Application to Intelligent Agents

    H. Katagiri, K. Hirasawa, J. Hu

    IEEE SMC 2000   Nashvile   3829 - 3834  2000

  • A New Method Based on Determining Error Surface for Designing Three Layer Neural Networks

    B. Lu, J. Murata, K. Hirasawa

    計測自動制御学会論文集   36 ( 7 ) 589 - 598  2000

     View Summary

    A method is proposed for designing three layer neural networks that gives relevant network structures which assure global minimization of learning errors for small training sets and small learning errors for big training sets both irrespective of the initial values. A condition on network structure is considered to achieve the above purpose, and a number of possible structures are provided together with their learning algorithms. Also, the generalization abilities of the network structures are discussed to guide the choice of structures in practice. All of the proposed structures for small training sets have zero errors after learning by a gradient-based algorithm and thus solve the local minima problem. The difference between them is in the level of locality and generalization abilities. For a big training set, first, the structure with zero learning errors for part of training data is obtained, then all of training data are used to train the network of the given structure, which improves the generalization abilities. Numerical examples are provided that support the present approach.

    DOI CiNii

  • RasID Training of Multi Agent Systems with Fuzzy Inference-Based Mutual Interactions

    K. Hirasawa, J. Hu, J. Misawa, J. Murata, H. Katagiri

    J. Machine Intelligence and Robotic Control   2 ( 1 ) 17 - 25  2000

  • Chaos control on universal learning networks

    Kotaro Hirasawa, Junichi Murata, Jinglu Hu, Chunzhi Jin

    IEEE Transactions on Systems, Man and Cybernetics Part C: Applications and Reviews   30 ( 1 ) 95 - 104  2000

     View Summary

    A new chaos control method is proposed which is useful for taking advantage of chaos and avoiding it. The proposed method is based on the following facts: 1) chaotic phenomena can be generated and eliminated by controlling maximum Lyapunov exponent of systems and 2) maximum Lyapunov exponent can be formulated and calculated by using higher order derivatives of Universal Learning Networks (ULN's). ULN's consist of a number of inter-connected nodes where the nodes may have any continuously differentiable nonlinear functions in them and each pair of nodes can be connected by multiple branches with arbitrary time delays. A generalized learning algorithm has been derived for the ULN's, in which both the first-order derivatives (gradients) and the higher order derivatives are incorporated. In simulations, parameters of ULN's with bounded node outputs are adjusted for maximum Lyapunov component to approach the target value. And, it has been shown that a fully connected ULN with three sigmoidal function nodes is able to generate and eliminate chaotic behaviors by adjusting the parameters.

    DOI CiNii

  • Universal learning network and its application to robust control

    Kotaro Hirasawa, Junichi Murata, Jinglu Hu, Chunzhi Jin

    IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics   30 ( 3 ) 419 - 430  2000

     View Summary

    Universal learning networks (ULN's) and robust control system design are discussed. ULN's provide a generalized framework to model and control complex systems. They consist of a number of inter-connected nodes where the nodes may have any continuously differentiable nonlinear functions in them and each pair of nodes can be connected by multiple branches with arbitrary time delays. Therefore, physical systems which can be described by differential or difference equations and also their controllers can be modeled in a unified way. So, ULN's constitute a superset of neural networks or fuzzy neural networks. In order to optimize the systems, a generalized learning algorithm is derived for the ULN's, in which both the first order derivatives (gradients) and the higher order derivatives are incorporated. The derivatives are calculated by using forward or backward propagation schemes. These algorithms for calculating the derivatives are extended versions of back propagation through time (BPTT) and real time recurrent learning (RTRL) by Williams in the sense that generalized nonlinear functions and higher order derivatives are dealt with. As an application of ULN's, the higher order derivative, one of the distinguished features of ULN's, is applied to realizing a robust control system in this paper. In addition, it is shown that the higher order derivatives are effective tools to realize sophisticated control of nonlinear systems. Other features of ULN's such as multiple branches with arbitrary time delays and using a priori information will be discussed in other papers.

    DOI CiNii

  • LimNet-Flexible Learning Network Containing Linear Properties

    J. Hu, K. Hirasawa, K. Kumamaru

    J. Advanced Computational Intelligence   3 ( 4 ) 303 - 311  1999

  • On-line Identification of Furnace Parameters for Coal-Fired Boiler Control

    Y. Fukayama, K. Hirasawa, K. Shimohira

    J. Robotics and Mechatronics   6 ( 5 ) 374 - 379  1999

  • Control of Decentralized Systems Based on Nash Equilibrium Concept of Game Theory

    K. Hirasawa, J. Hu, Y. Yamamoto

    J. Advanced Computational Intelligence   3 ( 4 ) 321 - 319  1999

  • Probabilistic Learning-Network-Based Robust Control Scheme for Nonlinear Systems

    J. Hu, K. Hirasawa, J. Murata

    J. Advanced Computational Intelligence   3 ( 6 ) 485 - 490  1999

  • Neurofuzzy Approach to Fault Detection of Nonlinear Systems

    J. Hu, K. Hirasawa, K. Kumamaru

    J. Advanced Computational Intelligence   3 ( 6 ) 524 - 531  1999

  • LimNet-Flexible Learning Network Containing Linear Properties

    J. Hu, K. Hirasawa, K. Kumamaru

    J. Advanced Computational Intelligence   3 ( 4 ) 303 - 311  1999

  • On-line Identification of Furnace Parameters for Coal-Fired Boiler Control

    Y. Fukayama, K. Hirasawa, K. Shimohira

    J. Robotics and Mechatronics   6 ( 5 ) 374 - 379  1999

  • Control of Decentralized Systems Based on Nash Equilibrium Concept of Game Theory

    K. Hirasawa, J. Hu, Y. Yamamoto

    J. Advanced Computational Intelligence   3 ( 4 ) 321 - 319  1999

  • Probabilistic Learning-Network-Based Robust Control Scheme for Nonlinear Systems

    J. Hu, K. Hirasawa, J. Murata

    J. Advanced Computational Intelligence   3 ( 6 ) 485 - 490  1999

  • Neurofuzzy Approach to Fault Detection of Nonlinear Systems

    J. Hu, K. Hirasawa, K. Kumamaru

    J. Advanced Computational Intelligence   3 ( 6 ) 524 - 531  1999

  • Learning Petri Network and its application to nonlinear system control

    K Hirasawa, M Ohbayashi, S Sakai, JL Hu

    IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART B-CYBERNETICS   28 ( 6 ) 781 - 789  1998.12

     View Summary

    According to recent knowledge of brain science, it is suggested that there exists functions distribution, which means that specific parts exist in the brain for realizing specific functions. This paper introduces a new brain-like model called Learning Petri Network (LPN) that has the capability of functions distribution and learning, The idea is to use Petri net to realize the functions distribution and to incorporate the learning and representing ability of neural network into the Petri net. The obtained LPN can be used in the same way as a neural network to model and control dynamic systems, while it is distinctive to a neural network in that it has the capability of functions distribution, An application of the LPN to nonlinear crane control systems is discussed. It is shown via numerical simulations that the proposed LPN controller has superior performance to the commonly-used neural network one.

    DOI CiNii

  • Learning Petri Network and its application to nonlinear system control

    K Hirasawa, M Ohbayashi, S Sakai, JL Hu

    IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART B-CYBERNETICS   28 ( 6 ) 781 - 789  1998.12

     View Summary

    According to recent knowledge of brain science, it is suggested that there exists functions distribution, which means that specific parts exist in the brain for realizing specific functions. This paper introduces a new brain-like model called Learning Petri Network (LPN) that has the capability of functions distribution and learning, The idea is to use Petri net to realize the functions distribution and to incorporate the learning and representing ability of neural network into the Petri net. The obtained LPN can be used in the same way as a neural network to model and control dynamic systems, while it is distinctive to a neural network in that it has the capability of functions distribution, An application of the LPN to nonlinear crane control systems is discussed. It is shown via numerical simulations that the proposed LPN controller has superior performance to the commonly-used neural network one.

    DOI CiNii

  • Computing Higher Order Derivatives in Universal Learning Networks

    K. Hirasawa, J. Hu, M. Ohbayashi, J. Murata

    J. Advanced Computational Intelligence   2 ( 2 ) 47 - 53  1998

  • Chaos Universal Learning Network Clustering Control

    K. Hirasawa, J. Misawa, J. Hu, J. Murata, M. Ohbayashi, T. Eki

    J. Robotics and Mechatronics   10 ( 4 ) 305 - 310  1998

  • RasID-Random Search for Neural Network Training

    J. Hu, K. Hirasawa, J. Murata

    J. Advanced Computational Intelligence   2 ( 4 ) 134 - 141  1998

  • Computing Higher Order Derivatives in Universal Learning Networks

    K. Hirasawa, J. Hu, M. Ohbayashi, J. Murata

    J. Advanced Computational Intelligence   2 ( 2 ) 47 - 53  1998

  • Chaos Universal Learning Network Clustering Control

    K. Hirasawa, J. Misawa, J. Hu, J. Murata, M. Ohbayashi, T. Eki

    J. Robotics and Mechatronics   10 ( 4 ) 305 - 310  1998

  • RasID-Random Search for Neural Network Training

    J. Hu, K. Hirasawa, J. Murata

    J. Advanced Computational Intelligence   2 ( 4 ) 134 - 141  1998

  • Energy Saving Elevator Group Supervisory Control System with Idle Cage Assignment using Genetic Network Programming

    Tiantian Zhang, Shingo Mabu, Lu Yu, Jin Zhou, Xianchao Zhang, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   994 - 999

  • Efficient Program Generation by Evolving Graph Structure with Multi-Start Nodes

    S. Mabu, K. Hirasawa

    Applied Soft Computing  

    DOI

  • Energy Saving Elevator Group Supervisory Control System with Idle Cage Assignment using Genetic Network Programming

    Tiantian Zhang, Shingo Mabu, Lu Yu, Jin Zhou, Xianchao Zhang, Kotaro Hirasawa

    ICROS-SICE International Joint Conference 2009   Fukuoka   994 - 999

  • Efficient Program Generation by Evolving Graph Structure with Multi-Start Nodes

    S. Mabu, K. Hirasawa

    Applied Soft Computing  

    DOI

▼display all

 

Internal Special Research Projects

  • 構成論的脳モデルに関する研究

    2005  

     View Summary

    研究室では、遺伝的アルゴリズム(GA)や遺伝的プログラミング(GP)とは異なる新しい進化論的計算手法ー遺伝的ネットワークプログラミング(Genetic Network Programming,GNP)の研究・開発を進めている。GNPは遺伝子として、判定ノードと処理ノードからなる有向グラフで構成されており、ノード間の接続およびノード関数を進化論的計算手法により最適化する点が特徴である。したがって、・必要な判定のみで適切な処理を実行する部分観測マルコフ過程に応用できる。・判定ノードと処理ノードの重複活用が可能なため、コンパクトな遺伝子を構成できる。・有向グラフによるネットワーク遺伝子のため、過去の状況に依存した判定が可能であり、 遺伝子の中にメモリー機能を持つことができる。など、従来の進化論的計算手法に比較して数多くの有利な点がある。本特定課題では、ロボットあるいはエージェントの人口脳をGNPを活用して構成論的に構築し、それを各種の応用に展開することにある。人口脳の構成に関しては・学習と進化を統合した最適化のアルゴリズム・脳の機能局在を実現するアルゴリズムなどを検討した。また人口脳の応用に関しては・エレベータ群管理システムのコントローラ・株の売買を行うトレーダの人口脳などに展開する研究を進めてきた。

  • 遺伝的ネットワークプログラミングを用いた脳のモデルに関する研究

    2004  

     View Summary

     遺伝的アルゴリズム(GA)や遺伝的プログラミング(GP)とは異なり、有向グラフ遺伝子を持つ遺伝的ネットワークプログラミング(Genetic Network Programming GNP )を開発し、これを利用した構成論的な人工脳を進化により構成する研究を行なった。GNPは判定ノードと処理ノードがネットワーク状に結合しており、判定ノードでは環境からの情報を判定し多分岐し、処理ノードでは環境に対して出力を行なうノードである。GNPの特徴は、「1」有限状態機械と異なり必要な情報を必要なときに取り込むことが出来るため、 部分マルコフ決定過程のプロセスをモデル化できる「2」判定ノードおよび処理ノードを重複して活用できるため、コンパクトな構成が可能になる「3」有向グラフによるネットワーク構成のため、GNPの内部に過去のノード遷移の履歴を 記憶することができる等である。GNPを人口脳として展開するために、GNPのアーキテクチャーの研究とその応用研究を行った。・GNPのアーキテクチャーの研究 機能局在型GNPの構成:大規模な仕事を複数個のサブの仕事に分割し、サブの仕事に対応する複数個のサブGNPで構成されるGNP 学習・進化型GNPの構成:GNPノードのサブノードの選択を学習により行い、ノードの接続を進化により行うGNP などの研究を行った。・GNPの応用の研究 上記のGNPを、ロボカップゲームのプレイヤ、エレベータ群管理のコントローラ、 株価予測および売買を行うディーラー、データベースの興味ある相関ルールの抽出機 などに展開する研究を行った。