HomePage of Kazuho Watanabe

[Japanese|Russian]

This is my previous homepage.
I'm going to move to Toyohashi University of Technology (Apr, 2014).

CV, Publications

--News--------------------------------------

  • Andrew Barron, Teemu Roos and Kazuho Watanabe:
    "Bayesian Properties of Normalized Maximum Likelihood and its Fast Computation"
    arXiv:1401.7116.
  • ACML2013 (Canberra, Australia, Nov, 13-15)
    Kazuho Watanabe, Teemu Roos and Petri Myllymäki:
    "Achievability of Asymptotic Minimax Regret in Online and Batch Prediction"<pdf>.
  • ICONIP2013 (Daegu, Korea, Nov, 3-7)
    "Vector Quantization Using Mixture of Epsilon-Insensitive Components"
    Special Session: Extracting Latent Features from Multi-Dimensional Data
  • --------------------------------------------

    --Past Topics--------------------------------

  • I visited HIIT (Sep, 16-27).

  • IEEE ITW2013 (Sevilla, Spain, Sep, 9-13)
    "Rate-Distortion Bounds for an Epsilon-Insensitive Distortion Measure"
  • WITMSE2013 (The University of Tokyo, Japan, Aug, 26-29)
    Kazuho Watanabe, Teemu Roos and Petri Myllymäki:
    "Achievability of Asymptotic Minimax Optimality in Online and Batch Coding"
  • The following paper has been accepted for IEEE ISIT2013 (Istanbul, Turkey, Jul, 7-12).
    Kazuho Watanabe, Shiro Ikeda:
    "Rate-Distortion Function for Gamma Sources under Absolute-Log Distortion Measure"
  • Technical meeting of IEICE, IBISML (Nagoya Institute of Technology, Japan, Mar, 4-5).
    Kazuho Watanabe, Teemu Roos and Petri Myllymäki:
    "Non-Achievability of Asymptotic Minimax Regret without Knowledge of the Sample Size"
  • arXiv Submission: CoRR abs/1302.6315
    "Rate-Distortion Bounds for an Epsilon-Insensitive Distortion Measure"
  • Mr. Miyamoto, a Ph.D. candidate in our lab, made a presentation about our recent work at APSIPA ASC 2012(Hollywood, California, USA, Dec, 3-6):
    A.Miyamoto, K.Watanabe, K.Ikeda, "Packet Loss Rate Estimation with Active and Passive Measurements"

  • In October 2012, I visited the Complex Systems Computation group in Helsinki Institute for Information Technology.

  • I participated in WITMSE2012 (Amsterdam, Netherlands, Aug, 27-30).
    K.Watanabe, S.Ikeda, "Convex formulation for nonparametric estimation of mixing distribution,"
    Proc. of the 2012 Workshop on Information Theoretic Methods in Science and Engineering, 2012. .<pdf>
  • I made a presentation at IMS-APRM (Tsukuba, Japan, Jul, 2-4).
    TCP08-2: "Approximate Bayesian Learning Using Bregman Divergence"<ppt>
  • The following paper has been published in Machine Learning Journal:
    "An alternative view of variational Bayes and asymptotic approximations of free energy".
    doi:10.1007/s10994-011-5264-5, <pdf>

  • An article entitled "Approximate Bayesian estimation of varying binomial process,"
    by K.Watanabe and M.Okada has been published in IEICE Transactions on Fundamentals.

  • I made a presentation at WITMSE2011(University of Helsinki, Finland, Aug,7-10).
    "Minimum variational stochastic complexity and average generalization error in latent variable models",
    Proc. of the 2011 Workshop on Information Theoretic Methods in Science and Engineering, 2011. <pdf>.

  • The following paper has been accepted:
    Kazuho Watanabe, Masato Okada, Kazushi Ikeda:
    ``Divergence Measures and a General Framework for Local Variational Approximation'', Neural Networks.
    doi:10.1016/j.neunet.2011.06.004, <pdf>
    This paper describes a general framework for the local variational approximation of Bayesian learning
    and explains its principle by using the Bregman divergence.

  • I attended the workshop WITMSE 2010 (Tampere University of Technology, Finland, Aug.16-18).
    "An alternative view of variational Bayes and minimum variational stochastic complexity",
    Proc. of the 2010 Workshop on Information Theoretic Methods in Science and Engineering, 2010. <pdf>

  • Our paper,
    "Simultaneous Clustering and Dimensionality Reduction Using Variational Bayesian Mixture Model", <doi>
    presented at IFCS/GfKl 2009 Conference, was selected for the "Best Paper Method Award 2009" of the GfKl (The German Classification Society).
    I presented a brief overview of this study in the GfKl 2010 symposium held at Karlsruhe Institute of Technology (KIT, Germany, Jul.21-23).

  • I attended the workshop on the state-space analysis(Kyoto Univ. Apr. 8-9th)
    and listened to some lectures about the state-space analyses especially in neuroscience applications.

  • "Variational Bayesian Mixture Model on a Subspace of Exponential Family Distributions"
    by K.Watanabe, S.Akaho, S.Omachi, and M.Okada
    has been published in IEEE Transactions on Neural Networks, vol.20,no11, pp.1783--1796, 2009. <link>.
    --------------------------------------------

    My former pages:
    The University of Tokyo(when I was a postdoc)
    Tokyo Institute of Technology(when I was a graduate student)

    Back to Mathematical Informatics Lab.mi@
    Why don't you have a NAIST time?