機械学習における情報幾何学的視点

Dates: 3-5, December 2014
Place: 3rd floor, Seminar room, Ikenohata Building, RIKEN BSI (理化学研究所和光キャンパス,脳科学総合研究センター池之端棟、3階セミナー室)
Registration: Registration is required. Please visit the left index "registration." (The number of participants reached the capacity of the seminar room. We have closed registration.) Free for participation.

Aim and Scope:
Information geometry started its history in 1980's. This framework has its mathematical background in Riemannian geometry and affine  geometry, and has been applied to many fields including statistics, control theory, signal processing, optimization theory, and information theory. Information geometry gave better understanding of the problems in those fields and provided some clues for improvements.

Machine learning is a collection of theories and methodologies developed in order to deal with new types of data. Each theory and methodology has its background in a related field, such as, statistics, computer science and optimization theory, where information geometry have made a lot of contributions. For further improvements of machine learning methods, information geometrical viewpoint will be useful. In this workshop, recent results of information geometry and machine learning will be reviewed, and future direction will be discussed.

Confirmed speakers:
  • Shun-ichi Amari, RIKEN, Japan
  • Nihat Ay, Max Planck Institute for Mathematics in the Sciences, Germany
  • Shinto Eguchi,  The Institute of Statistical Mathematics, Japan
  • Kenji Fukumizu,  The Institute of Statistical Mathematics, Japan
  • Mark Girolami,  University of Warwick, UK
  • Masayuki Henmi,  The Institute of Statistical Mathematics, Japan
  • Shiro Ikeda,  The Institute of Statistical Mathematics, Japan
  • Takafumi Kanamori, Nagoya University, Japan
  • Fumiyasu Komaki,  University of Tokyo, Japan
  • Hiroshi Matsuzoe,  Nagoya Institute of Technology, Japan
  • Guido Montufar, Max Planck Institute for Mathematics in the Sciences, Germany
  • Frank Nielsen,  Ecole Polytechnique, France
  • Atsumi Ohara,  University of Fukui, Japan
  • Takashi Tsuchiya, National Graduate Institute for Policy Research, Japan
(All speakers are invited. The speaker list has been changed on 15 Oct 2014.)

Organizers:
  • Shiro Ikeda,  The Institute of Statistical Mathematics, Japan
  • Kenji Fukumizu,  The Institute of Statistical Mathematics, Japan
Contact
 Shiro Ikeda,  shiro [atmark] ism [dot] ac [dot] jp