Presentation 1998/2/6
Dynamic function approximation by online EM algorithm
Shin Ishii, Masa-aki Sato,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this research report, we propose an online EM algorithm for Normalized Gaussian Network which is composed of normalized Gaussian functions with linear coefficients. A regularization method to deal with singular input distribution, and unit creation-deletion mechanisms are also discussed. Our approach is suitable for dynamical environments where the input-output distribution of data changes in time. A computer simulation is done for a reinforcement learning problem; the result is encouraging.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) EM algorithm / Online learning / Radial basis functions / Stochastic model / Dynamic function approximation / Reinforcement learning
Paper #
Date of Issue

Conference Information
Committee NLP
Conference Date 1998/2/6(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Nonlinear Problems (NLP)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Dynamic function approximation by online EM algorithm
Sub Title (in English)
Keyword(1) EM algorithm
Keyword(2) Online learning
Keyword(3) Radial basis functions
Keyword(4) Stochastic model
Keyword(5) Dynamic function approximation
Keyword(6) Reinforcement learning
1st Author's Name Shin Ishii
1st Author's Affiliation Nara Institute of Science and Technology()
2nd Author's Name Masa-aki Sato
2nd Author's Affiliation ATR Human Information Processing Research Laboratories
Date 1998/2/6
Paper #
Volume (vol) vol.97
Number (no) 531
Page pp.pp.-
#Pages 8
Date of Issue