KAMATANI, Kengo/ 鎌谷 研吾

Associate Professor
Graduate School of Engineering Science, Osaka University
kamatani at sigmath.es.osaka-u.ac.jp

1-3 Machikaneyama-cho, Toyonaka, Osaka, Japan








Information

  • 2016.Apr.13: Pseudo-code of MpCN algorithm, which is a derivative-free general purpose MCMC, can be found here. See this paper for high-dimension asymptitics, and see this recent paper for ergodicity. The latter paper is short and is much easier than the former!
  • 2015.Sep.16: CREST team website is here. Research Director is Prof. Nakahiro Yoshida in Tokyo. This project aims at developing mathematical statistics and probability theory to provide methodologies for modeling and analysis of complex random systems.
  • 2015.Sep.16: Check probability and statistics seminars in Osaka University; here, here, here and here.


Research Interest

  • Bayesian computation: Markov chain Monte Carlo, Sequential Monte Carlo and other related methods
  • Asymptotic theory: Asymptotic theory for Bayesian inference, Asymptotic properties of Hidden Markov model
  • Statistical Genetics: Statistical approach for genetics data with latent structure


Education

2003 B.A. Mathematics, Unviersity of Tokyo
2005 M.A. Mathematical Sciences, University of Tokyo
2008 Ph.D. Mathematical Sciences, University of Tokyo


Publication



Talks and forthcoming events

2016
  • Kyusyu, Nov
  • Markov chain Monte Carlo for high-dimensional target distribution, Stochastic Analysis and Statistics 4, Komaba, Oct 31-Nov 01
  • マルコフ連鎖モンテカルロ法の高次元解析,JJS, Kanazawa, Sep 5-7
  • Some properties of the mixed preconditioned Crank-Nicolson algorithm, IMS-APRM, Hong Kong, China, June, 27-30
  • Reversible proposal MCMC in high dimension, SIAM Conference on Uncertainty Quantification (UQ16), Lausanne, Switzerland, Apr, 5-8
  • High-dimensional asymptotic properties of Markov chain Monte Carlo methods for heavy-tailed target distributions, Paris, Mar 23
  • マルコフ連鎖のエルゴード性とregular variation, 日本統計学会春季集会, Tohoku, Mar 5
  • Ookayama, Tokyo, Feb
  • Ergodicity of MpCN and related MCMC algorithms, Asymptotic Statistics and Computations, Komaba, Tokyo, Feb 15
  • Markov chain Monte Carlo in high-dimension with heavy-tailed target probability distributions, MCMski V, Lenzerheide, Switzerland, Jan, 5-7
2015
  • Efficient strategy for the Markov chain Monte Carlo in high-dimension and its implementation, IASC-ARS, Singapore, Dec
  • On some ergodic properties of the MpCN algorithm, ERCIM, London, Dec
  • マルコフ連鎖モンテカルロ法の高次元漸近論, 大規模複雑データの理論と方法論:最前線の動向, Tsukuba, Nov
  • 対称な提案を有するMCMCの高次元解析, Tokyo, Oct
  • 高次元でのマルコフ連鎖モンテカルロ法, CREST Meeting, Tokyo, Sep
  • Asymptotic theory of Markov chain Monte Carlo method in high-dimension, MSJ, Kyoto, Sep
  • 高次元での高速なマルコフ連鎖モンテカルロ法, JSS, Okayama, Sep
  • Efficient strategy of MCMC in high-dimension and its application to diffusion processes, SAPS-X, LeMans, France, Mar [ PDF ]
  • Efficient strategy for the Markov chain Monte Carlo in high-dimension with heavy-tailed target probability distribution, NUS, Singapore, Feb
  • Efficient construction of MCMC in high-dimension, 大規模統計モデリングと計算統計, Komaba, Feb
2014
  • Hybrid multi-step estimators for stochastic differential equations based on sampled data, ERCIM, Pisa, Italy, Dec
  • Rate optimality of Random walk Metropolis algorithm in high-dimension with heavy-tailed target distribution, Niigata
  • マルコフ連鎖モンテカルロ法の退化性の評価, Tokei-kanren gakkai, Tokyo, Sep
  • Rate optimality of Random walk Metropolis algorithm in high-dimension with heavy-tailed target distribution, DynStoch 2014, Warwick, UK, Sep
  • The order of degeneracy of Markov chain Monte Carlo Method, IMS-APRM 2014, Taipei, Chinese Taipei, 1st, Jul
  • Rate optimality of Markov chain Monte Carlo in high-dimension, Osaka, Apr
  • Weak consistency for Metropolis-Hastings algorithm in high-dimension, Komaba, Mar
  • マルコフ連鎖モンテカルロ法の拡散過程への収束, Kunitachi, Feb
  • (Poster) Local consistency of Markov chain Monte Carlo with some applications, MCMSki 2014, Chamonix, France, Jan
2013
  • マルコフ連鎖モンテカルロ法の一致性, Kyusyu, Nov
  • 混合モデルのマルコフ連鎖モンテカルロ法の漸近理論, Tokei-kanren gakkai,Osaka, Sep
  • 混合モデルのベイズ推定, Summer seminar, Hiroshima, Aug
  • Various order of degeneracies of Markov chain Monte Carlo for categorical data, EMS2013, Budapest, 20-25th, July
  • Order of degeneracy of several MCMCs, Asymptotic Statistics and Computations 2013, Komaba, 27-28th, Mar
  • Non degenerate MCMC for categorical model, Asymptotic Expansions for Various Models and Their Related Topics, RIMS, Kyoto, 4-6th, Mar
2012
  • Contiguous proposal for Metropolis-Hastings algorithms, SART2012, Komaba, Tokyo, 19-22th Dec
  • Non-regular model for Bayesian computation, Takanawa, Tokyo, 18-19th Dec
  • (Poster) Efficient Monte Carlo Strategy for Simple Mixture Model, SuSTaIn, Bristol, UK, 25th, Sep
  • Local consistency of MCMC and its application to cumulative link model, BigMC, Institut Henri Poincare, Paris, France, 20th, Sep
  • Markov chain Monte Carlo methods for simple mixture model, IMS-APRM 2012, Tsukuba, July
  • Efficiency of Monte Carlo methods, Waseda University, Tokyo, 30th, June
  • (Poster) Asymptotic properties of Monte Carlo strategies for cumulative link model, ISBA 2012, Kyoto, 27th, June [ Abst ]
  • Weak consistency of Markov chain Monte Carlo, Seminar on Probability, Osaka, 8th, May [ Abst ]
  • ベイズ統計学におけるモンテカルロ法, Open Seminar of Data Science, Osaka, 25th Apr
  • Asymptotic properties of MCMC for cumulative link model, Seminar on Probability and Statistics, Komaba, Apr
2011
  • Local degeneracy of MCMC for cumulative logit model, SART2011, Komaba, Dec
  • マルコフ連鎖モンテカルロ法の退化性 (Degeneracy of Markov chain Monte Carlo methods), Statistics Summer Seminar, Suwa, Nagano, June
  • MCMCの収束の決定論的扱いと退化性 (Decision theoretic view of MCMC and its degeneracy), 数理統計学の新たな展開, Tsukuba, June
  • ベイズ統計学でのData Augmentationの手法 (Data augmentation strategy in Bayesian statistics), Emergent Dynamics in Nonlinear Science, Komaba, May
  • Weak Convergence of Markov chain Monte Carlo II, Asymptotic Statistics of Stochastic Processes - VIII, Le Mans, France, Mar [ PDF ]
  • Weak Convergence of Markov chain Monte Carlo, Statistical inference and numerical analysis of stochastic processes, Florence, Italy, Mar [ PDF ]
  • Convergence of MCMC for Simple Binomial Model, Statistics for Stochastic Processes, Komaba, Tokyo, Feb
2010
  • MCMC法の解析における漸近的方法, RIMS, Kyoto, Nov
  • Convergence of Markov measure valued random variables and its application to MCMC, SPA 2010, Senri life science center, Osaka, Sep
  • マルコフチェインモンテカルロ法のエルゴード性の解析, マクロ経済動学の非線形数理, RIMS, Kyoto, Sep
  • マルコフチェインモンテカルロ法の弱収束, Toukei Kanren Gakkai Rengo Taikai, Waseda University, Tokyo, Sep
  • Metropolis-Hastings Algorithm for Mixture Model and its Weak Convergence, COMPSTAT 2010, CNAM, Paris, Aug [ PDF ]
  • Validity of the EM algorithm for haplotype frequency estimation, Summer Seminar, Izu, Shizuoka, Aug
  • Weak convergence of Markov chain Monte Carlo method and its application to Yuima, Seminar on Probability and Statistics, UT, Komaba, Tokyo, Jun
  • Weak convergence of the Gibbs sampler and its application to mixture model, Stochastic Analysis of the Advanced Statistical Models, Hiroshima University, Hiroshima, Mar
  • Weak convergence of the Gibbs sampler for some simple models, Stochastic Analysis and Statistical Inference V, UT, Komaba, Tokyo, Feb
  • Weak Convergence of the Gibbs sampler, Seminar on Statistics, UT, Hongo, Tokyo, Jan
2009
  • Non-regular Behaviors of Monte Carlo Methods for finite Mixture Models, Theory and Applications on Statistical Inference and Probability Analysis and Related Areas, Akita city, Akita, Dec
  • Irregular Behaviors of the Gibbs Sampler for the Mixture Model with Strong Identifiability Conditions, Recent Results on Mathematical Statistics and Related Areas, Takasaki city, Gunma, Nov
  • Asymptotic behavior of the Gibbs sampler, Statistics Young Summer Seminar, Fukui city, Fukui, Aug
  • Some Non-regular Models for the EM Algorithm and the Gibbs Sampler, Stochastic Analysis and Statistical Inference VI, UT, Komaba, Tokyo, Feb
2008
  • On some asymptotic properties of the Gibbs sampler, CASTA2008, Kyoto University, Kyoto, Dec
  • Convergence properties of the Gibbs sampler and related algorithms, Stochastic Analysis and Statistical Inference III, Komaba, Nov
  • On some asymptotic properties of the EM algorithm, Stat. Sem., UT, Hongo, Oct
  • Large sample theory for the EM algorithm and the Gibbs sampling, Efficient Monte Carlo, Sonderborg, Denmark, July [ PDF ]
  • Asymptotic behaviors of the Gibbs sampling, Stochastic Analysis and Statistical Inference II, Komaba, Feb
2007
  • Asymptotic Statistics for Haplotype Association Study, Problems on Statistical Decision Theory, Hokkaido, Dec
  • Local properties for Markov chain Mote Carlo algorithm, Stochastic Analysis and Statistical Inference, Komaba, Nov
  • Haplotype Association Study: Approaches from MCMC, Stat. Sem., UT, Hongo, Nov
  • Haplotype Association Study: Approaches from EM algorithm, Sem. on Prob. & Stat., UT, Komaba, Nov
2006
  • A Note on Haplotype estimation, Math. Sci. on Stat. Model, ISM, Azabu, Nov
  • A Note on Haplotype estimation, Sem. on Prob. & Stat., UT, Komaba, Nov
  • A Note on Haplotype estimation, Stat. Sem., UT, Hongo, Nov
  • Central Limit Theorem for polynomial ergodic Markov chain, COE poster session, UT, Komaba, Sep
  • A Note on Haplotype Estimation, Stat. Summer Sem., Saitama, Aug
2005
  • MHs with high acceptance ratios, the 5th IASC Asian Conf. on Stat. Comput., Univ. of Hong Kong, China, Dec
  • MHs with acceptance ratios of nearly 1, Asymp. Methods for Prob. and Stat., Komaba, Dec
  • MHs with acceptance ratios of nearly 1, COE poster session, UT, Komaba, Sep
  • MHs with acceptance ratios of nearly 1, Stat. Summer Sem., Ohita, Aug
  • MH whose acceptance rate is almost 1, Stat. Sem., UT, Hongo, Jun
  • MH whose acceptance rate is almost 1, Sem. on Prob. & Stat., UT, Komaba, Jun
2004
  • Transformed MH with polynomial or geometrical ergodicity, Sem. on Prob. & Stat., UT, Komaba, Dec
  • Adapted MH using histogram estimate, Stat. Summer Sem., Wakayama, Aug
  • Adapted MH using histogram estimate, Stat. Sem., UT, Hongo, Jun
  • Markov chain in a general state space, Sem. on Prob. & Stat., UT, Komaba, Jan


MH: Metropolis-Hastings Algorithm, UT: University of Tokyo, COE: Centers of Excellence, ISM: Institute of Statistical Mathematics


Professional Experience

07 Apr- 08 Sep, JSPS Research Fellowships for Young Scientists (DC2)
08 Oct- 09 Mar, JSPS Research Fellowships for Young Scientists (PD)
09 Apr-11 Oct, Project Research Associate at Graduate School of Mathematical Sciences, University of Tokyo
11 Oct-14 Mar, Assistant Professor at Graduate School of Engineering Science, Osaka University
14 Apr- Associate Professor at Graduate School of Engineering Science, Osaka University

Lecture
10 Summer, 確率モデルと統計手法演習 (Univ. of Tokyo)
11 Summer, 確率モデルと統計手法演習 (Univ. of Tokyo)
11 Winter, Statistics C2, 数学IB演習 (Osaka Univ.)
12 Summer, 計算数理A (Osaka Univ.)
12 Winter, Statistics C2 (Osaka Univ.)
13 Summer, 計算数理A (Osaka Univ.)
13 Winter, Statistics C2 (Osaka Univ.)
14 Summer, 計算数理A (Osaka Univ.)
14 Winter, Statistics C2, Time series analyis, Mathematics C (Osaka Univ.)


Teaching Assistant
04 Summer, Teaching Asistant, Computing Mathematics for the third year students
05 Summer, Teaching Assitant, Mathematics I(A) for the first year students
06 Summer, Teaching Assitant, Mathematics I(A) for the first year students
06 Winter, Teaching Assitant, Probability and Statistics for the third year students

Research Assistant
05 Apr- 06 Mar, Research Assistant, supported by COE
06 Apr- 07 Mar, Research Assistant, supported by COE

Other activities
  • 05 Apr- now, one of the organizers of Seminar on Probability and Statistics
  • Yuima Project Developer. Yuima is a project for simulation and inference of multidimensional stochastic differential equations in R.
  • 確率の応用例,東京大学高校生のための現代数学講座 (lectures for highschool students in Gunma pref.), Tambara seminar house, Gunma, Jul
  • One of the organizers of Statistics Young Summer Seminar 2012
  • 君は現代の予言者になれるか, 2016年度 東京大学高校生のための現代数学講座, テーマ「確率と統計」, 玉原


Fellowships and Financial Aid
  • Best Student Paper Award and Wakimoto Memorial Fund at the 5th IASC Asian Conf. on Stat. Comput., 2005
  • JSPS Research Fellowships for Young Scientists (DC2), 2007 Apr -2009 Mar
  • Grant-in-Aid for Scientific Research from the Ministry of Education, Japan, (Grant-in-Aid for Young Scientists (B) 22740055) 2010-2011
  • Grant-in-Aid for Scientific Research from the Ministry of Education, Japan, (Grant-in-Aid for Young Scientists (B) 24740062) 2012--2015
  • Grant-in-Aid for Scientific Research from the Ministry of Education, Japan, (Grant-in-Aid for Scientific Research (C) --) 2016--
  • Ogawa Prize 2014, Japan Statistical Society
  • Osaka Group, Mathematical statistics and stochastic analysis for modeling and analysis of complex random systems, JST CREST (Research Director: Nakahiro Yoshida, Univ. of Tokyo)
JSPS: Japan Society for the Promotion of Science




KAMATANI, Kengo WEB