Updated on 2025/07/30

写真a

 
WOLFER, Geoffrey
 
Affiliation
Affiliated organization, Center for Data Science
Job title
Assistant Professor(without tenure)
Degree
Doctor of Philosophy in Computer Science ( 2020.10 Ben-Gurion University of the Negev )
Mail Address
メールアドレス
Homepage URL
https://geo-wolfer.gitlab.io

Research Experience

  • 2024.04
    -
    Now

    Waseda University   Center for Data Science   Assistant Professor

  • 2022.03
    -
    2024.03

    RIKEN   Center for Advanced Intelligence Project (AIP)   Special Postdoctoral Researcher (SPDR)

  • 2020.12
    -
    2022.03

    Tokyo University of Agriculture and Technology   Institute of Engineering Division of Advanced Information Technology & Computer Science   JSPS Postdoctoral Fellow

  • 2024.08
    -
    Now

    RIKEN   Center for Advanced Intelligence Project (AIP)   Visiting Scientist

  • 2013.11
    -
    2016.09

    Dalet S.A.   R&D Engineer

  • 2013.09
    -
    2016.08

    University of Strasbourg   Adjunct Lecturer (Enseignant Vacataire)

    Distance education

▼display all

Education Background

  • 2016.10
    -
    2020.10

    Ben-Gurion University of the Negev   Department of Computer Science  

    Doctor of Philosophy

  • 2011.09
    -
    2013.09

    Keio University   Graduate School of Science and Technology   Science for Open and Environmental Systems  

    Master of Science in Engineering

  • 2009.09
    -
    2011.08

    Ecole Centrale de Nantes  

    Engineer's Degree (Diplôme d'Ingénieur)

  • 2007.09
    -
    2009.08

    Lycée Kléber   Higher School Preparatory Classes (CPGE)   MPSI・MP・Computer Science  

Committee Memberships

  • 2025
    -
    　

    International Conference on Machine Learning (ICML'25)  Meta Reviewer

  • 2025
    -
    　

    International Conference on Artificial Intelligence and Statistics (AISTATS'25)  Area Chair

  • 2024
    -
    　

    Conference on Learning Theory (COLT'24)  Program Committee

  • 2024
    -
    　

    International Conference on Artificial Intelligence and Statistics (AISTATS'24)  Area Chair

  • 2023
    -
    　

    Conference on Learning Theory (COLT'23)  Program Committee

  • 2023
    -
    　

    International Conference on Artificial Intelligence and Statistics (AISTATS'23)  Area Chair

  • 2022
    -
    　

    Conference on Learning Theory (COLT'22)  Program Committee

  • 2021
    -
    　

    Conference on Learning Theory (COLT'21)  Program Committee

▼display all

Research Areas

  • Statistical science   PAC framework, minimax theory, theoretical machine learning, distribution testing / Theory of informatics   Information geometry / Applied mathematics and statistics   Mathematical statistics, Markov chains, concentration of measure

Research Interests

  • Information Geometry

  • Applied probability

  • Concentration of measure

  • Theoretical machine learning

  • Mathematical statistics

  • Markov chains

▼display all

Awards

  • Outstanding Meta Reviewer

    2025.05   International Conference on Machine Learning (ICML'25)  

  • Best poster

    2021.08   Croucher Summer Course in Information Theory (CSCIT'21)  

  • Best student paper award, honorable mention

    2020.02   International Conference on Algorithmic Learning Theory (ALT'20)   Mixing Time Estimation in Ergodic Markov Chains from a Single Trajectory with Contraction Methods

 

Papers

  • Variance-Aware Estimation of Kernel Mean Embedding

    Geoffrey Wolfer, Pierre Alquier

    Journal of Machine Learning Research   26 ( 57 ) 1 - 48  2025.04  [Refereed]

    Authorship：Lead author, Corresponding author

  • Improved Estimation of Relaxation Time in Nonreversible Markov Chains

    Geoffrey Wolfer, Aryeh Kontorovich

    The Annals of Applied Probability   34 ( 1A ) 249 - 276  2024.02  [Refereed]

    Authorship：Lead author, Corresponding author

    DOI

  • Information Geometry of Reversible Markov Chains

    Geoffrey Wolfer, Shun Watanabe

    Information Geometry   4 ( 2 ) 393 - 433  2021.11  [Refereed]

    Authorship：Lead author, Corresponding author

  • Statistical Estimation of Ergodic Markov Chain Kernel over Discrete State Space

    Geoffrey Wolfer, Aryeh Kontorovich

    Bernoulli   27 ( 1 ) 532 - 553  2021.02  [Refereed]  [International journal]  [International coauthorship]

    Authorship：Lead author, Corresponding author

    DOI

    Scopus

    10
    Citation
    (Scopus)

  • Mixing Time Estimation in Reversible Markov Chains from a Single Sample Path

    Daniel Hsu, Aryeh Kontorovich, David A. Levin, Yuval Peres, Csaba Szepesvári, Geoffrey Wolfer

    The Annals of Applied Probability   29 ( 4 ) 2439 - 2480  2019.08  [Refereed]  [International journal]  [International coauthorship]

    DOI

    Scopus

    21
    Citation
    (Scopus)

  • Combinatorial Characterization of Exponential Families of Lumpable Stochastic Matrices

    Shun Watanabe, Geoffrey Wolfer

    arXiv:2412.08400    2024.12

    Authorship：Corresponding author

  • Empirical and Instance‐Dependent Estimation of Markov Chain and Mixing Time

    Geoffrey Wolfer

    Scandinavian Journal of Statistics   51 ( 2 ) 557 - 589  2024.06  [Refereed]

    Authorship：Lead author, Corresponding author

    　View Summary

    Abstract

    We address the problem of estimating the mixing time of a Markov chain from a single trajectory of observations. Unlike most previous works which employed Hilbert space methods to estimate spectral gaps, we opt for an approach based on contraction with respect to total variation. Specifically, we estimate the contraction coefficient introduced in Wolfer (2020), inspired from Dobrushin's. This quantity, unlike the spectral gap, controls the mixing time up to strong universal constants and remains applicable to nonreversible chains. We improve existing fully data‐dependent confidence intervals around this contraction coefficient, which are both easier to compute and thinner than spectral counterparts. Furthermore, we introduce a novel analysis beyond the worst‐case scenario by leveraging additional information about the transition matrix. This allows us to derive instance‐dependent rates for estimating the matrix with respect to the induced uniform norm, and some of its mixing properties.

    DOI

    Scopus

  • Geometric Aspects of Data-Processing of Markov Chains

    Geoffrey Wolfer, Shun Watanabe

    Transactions of Mathematics and Its Applications   8 ( 1 )  2024.05  [Refereed]

    Authorship：Lead author, Corresponding author

    DOI

  • Systematic Approaches to Generate Reversiblizations of Markov Chains

    Michael C.H. Choi, Geoffrey Wolfer

    IEEE Transactions on Information Theory   70 ( 5 ) 3145 - 3161  2024.05  [Refereed]

    DOI

  • Optimistic Estimation of Convergence in Markov Chains with the Average-Mixing Time

    Geoffrey Wolfer, Pierre Alquier

    arXiv:2402.10506    2024.02

    Authorship：Lead author, Corresponding author

  • Geometric Reduction for Identity Testing of Reversible Markov Chains

    Geoffrey Wolfer, Shun Watanabe

    Lecture Notes in Computer Science   14071   328 - 337  2023.08  [Refereed]

    Authorship：Lead author, Corresponding author

    DOI

    Scopus

  • Information Geometry of Markov Kernels: a Survey

    Geoffrey Wolfer, Shun Watanabe

    Frontiers in Physics   11  2023.07  [Refereed]  [Invited]

    Authorship：Lead author, Corresponding author

    　View Summary

    Information geometry and Markov chains are two powerful tools used in modern fields such as finance, physics, computer science, and epidemiology. In this survey, we explore their intersection, focusing on the theoretical framework. We attempt to provide a self-contained treatment of the foundations without requiring a solid background in differential geometry. We present the core concepts of information geometry of Markov chains, including information projections and the pivotal information geometric construction of Nagaoka. We then delve into recent advances in the field, such as geometric structures arising from time reversibility, lumpability of Markov chains, or tree models. Finally, we highlight practical applications of this framework, such as parameter estimation, hypothesis testing, large deviation theory, and the maximum entropy principle.

    DOI

    Scopus

  • Learning and Identity Testing of Markov Chains

    Geoffrey Wolfer, Aryeh Kontorovich

    Handbook of Statistics   49   85 - 102  2023  [Refereed]  [Invited]

    Authorship：Lead author, Corresponding author

    DOI

    Scopus

  • Identity Testing of Reversible Markov Chains

    Sela Fried, Geoffrey Wolfer

    Proceedings of The 25th International Conference on Artificial Intelligence and Statistics (AISTATS'22)   151   798 - 817  2022.04  [Refereed]

    Authorship：Last author

  • Dimension-Free Empirical Entropy Estimation

    Doron Cohen, Aaron Koolyk, Aryeh Kontorovich, Geoffrey Wolfer

    Advances in Neural Information Processing Systems 34 (NeurIPS'21)    2021.12  [Refereed]

  • Learning Discrete Distributions with Infinite Support

    Doron Cohen, Aryeh Kontorovich, Geoffrey Wolfer

    Advances in Neural Information Processing Systems (NeurIPS'20)   33  2020.12  [Refereed]  [International journal]  [International coauthorship]

  • Minimax Testing of Identity to a Reference Ergodic Markov Chain

    Geoffrey Wolfer, Aryeh Kontorovich

    Proceedings of the 23rd International Conference on Artificial Intelligence and Statistics (AISTATS'20)   108   191 - 201  2020.08  [Refereed]  [International journal]  [International coauthorship]

    Authorship：Lead author, Corresponding author

  • Mixing Time Estimation in Ergodic Markov Chains from a Single Trajectory with Contraction Methods

    Geoffrey Wolfer

    Proceedings of the 31st International Conference on Algorithmic Learning Theory (ALT'20)   117   890 - 905  2020.02  [Refereed]  [International journal]  [International coauthorship]

    Authorship：Lead author, Corresponding author

    　View Summary

    Best student paper award, honorable mention.

  • Estimating the Mixing Time of Ergodic Markov Chains

    Geoffrey Wolfer, Aryeh Kontorovich

    Proceedings of the 32nd Conference on Learning Theory (COLT' 19)   99   3120 - 3159  2019.06  [Refereed]  [International journal]  [International coauthorship]

    Authorship：Lead author, Corresponding author

  • Minimax Learning of Ergodic Markov Chains

    Geoffrey Wolfer, Aryeh Kontorovich

    Proceedings of the 30th International Conference on Algorithmic Learning Theory (ALT'19)   98   904 - 930  2019.03  [Refereed]  [International journal]  [International coauthorship]

    Authorship：Lead author, Corresponding author

▼display all

Research Projects

  • Improving Information Geometry of Markov Chains for Data-Science

    Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research

    Project Year :

    2023.04
    -
    2026.03
     

  • Improve Theoretical Guarantees and Empirical Procedures for Assessment of Average Convergence in Markov Chains

    Waseda University 

    Project Year :

    2024.04
    -
    2025.03
     

  • New Interactions Between Statistics and Information Theory for Markov Chains

    Japan Society for the Promotion of Science  Grants-in-Aid for Scientific Research

    Project Year :

    2021.04
    -
    2022.03
     

 

Syllabus

▼display all

 

Sub-affiliation

  • Affiliated organization   Global Education Center

Internal Special Research Projects

  • Improve Theoretical Guarantees and Empirical Procedures for Assessment of Average Convergence in Markov Chains

    2024   Pierre Alquier

    　View Summary

    The convergence rate of a Markov chain towardsits stationary distribution is typically assessed using the concept of worst-casetotal variation *mixing time*. However, this quantity is pessimistic and is challengingto infer from a single stream of data. The goal of our research program, in collaborationwith Pierre Alquier (ESSEC Business School, Singapore) is to advocate for theuse of the average-mixing time as a more optimistic and demonstrablyeasier-to-estimate alternative.We had already demonstrated that estimatingthe average mixing time was both possible and beneficial in some settings andour plan was to improve our estimation procedures, both theoretically and in practice.Towards our goal, we have obtained thefollowing results in FY24:1/ We improved minimax ProbablyApproximately Correct (PAC) rates for estimating individual β-mixingcoefficients of discrete time, time homogeneous Markov chains over countablestate spaces by obtaining finite sample bounds in the *general ergodic* setting.Until now, rates in the general ergodic setting were only known for the MeanAbsolute Deviation (MAD), and PAC rates were only available in the *uniformlyergodic* setting, i.e. in the restrictive setting where the (worst-case) mixingtime is finite. Our new bound depends on the *average* mixing time instead, andthus remains valid even when the mixing time is infinite.2/ Building upon 1/ we obtained a PAC boundfor estimating the average mixing time from a single countable Markovian trajectoryin a *general ergodic setting*. Until now, only rates in the uniformly ergodicsetting were available.3/ Finally, we obtained estimation ratesfor Kernel Mean Embeddings (KME) with time-dependent data. Specifically, we extendedour previous estimation results from an iid setting to mixing processes byobtaining confidence intervals involving a covariance parameter in the ReproducingKernel Hilbert Space (RKHS) and mixing coefficients for both φ-mixing and β-mixing processes. In particular, we demonstrated that whenthe underlying process is a countable Markov chain, our framework based on theaverage mixing time is readily applicable.We presented our research results on the averagemixing time at the following venues:[1] IEEE East Asian School of InformationTheory 2024 (EASIT'24)Shonan, Kanagawa, July 30-August 2, 2024[2] Bernoulli-IMS 11th World Congress inProbability and Statistics (Bernoulli-IMS'24) Bochum, Germany, August 12-16, 2024[3] The 27th Information-Based InductionSciences Workshop (IBIS'24)Omiya, Saitama, November 4-7, 2024 [4] The 46th Symposium on Information Theoryand its Applications (SITA'24)Awara, Fukui, December 10-13, 2024