Centers & Labs

RIKEN Center for Advanced Intelligence Project

Geometric Learning Team

Team Leader: Takashi Takenouchi (Ph.D.)
Takashi  Takenouchi(Ph.D.)

Probabilistic models are widely applicable to represent various kinds of phenomena and data, and appropriate estimation of parameters is an important issue for precise inference. Our target is models for which conventional methods cannot be easily applicable, and our purpose is to construct a framework of estimation having “good” statistical properties, by utilizing local information of models and information geometric properties.

Main Research Field


Related Research Fields

Computer Science / Mathematics

Research Subjects

  • Statistical Machine Learning
  • Information Geometry

Selected Publications

Papers with an asterisk(*) are based on research conducted outside of RIKEN.
  1. *Takenouchi, T.:
    "A Novel Parameter estimation method for Boltzmann machine"
    Neural computation, 27(11), pp. 5673-5694 (2015).
  2. *Takenouchi, T., Komori O., and Eguchi S.:
    "Binary Classification with a Pseudo Exponential Model and Its Application for Multi-Task Learning"
    Entropy, 17, pp. 5673-5694 (2015).
  3. *Takenouchi, T., and Kanamor, T.:
    "Empirical Localization of Homogeneous Divergences on Discrete Sample Spaces"
    Neural Information Processing Systems (2015).
  4. *Takenouchi, T., Komori O., and Eguchi S.:
    "An extension of the Receiver Operating Characteristic curve and AUC-optimal classification"
    Neural computation, 24(10), pp. 2789-2824 (2012).
  5. *Takenouchi, T., and Ishii, S.:
    "Ternary Bradley-Terry model-based decoding for multi-class classification and its extensions"
    Machine Learning, 85(3), pp.249-272 (2011).
  6. *Takenouchi, T., and Ishii, S.:
    "A multi-class classification method based on decoding of binary classifiers"
    Neural Computation, 21(7), pp.2049-2081 (2009).
  7. *Takenouchi, T., Eguchi S., Murata, N. and Kanamor, T.:
    "Robust boosting algorithm against mislabeling in multi-class problems"
    Neural Computation, 20(16), pp.1596-1630 (2008).
  8. *Kanamor, T., Takenouchi, T., Eguchi, S., and Murata, N.:
    "Robust Loss Functions for Boosting"
    Neural Computation, 19, pp.2183-2244 (2007).
  9. *Murata, N., Takenouchi, T., Kanamor, T and Eguchi, S.:
    "Information geometry of U-Boost and Bregman divergence"
    Neural Computation, 16, pp.1437-1481 (2004).
  10. *Takenouchi, T., and Eguchi, S.:
    "Robustifying AdaBoost by adding the naive error rate"
    Neural Computation, 16, pp.767-787 (2004).