Presentation 2005/3/23
Bayesian noisy ICA for source switching environments
Junichiro HIRAYAMA, Shinichi MAEDA, Shin ISHII,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Most independent component analysis (ICA) algorithms assume that the number of sources is constant for all samples. This assumption does not necessarily hold, however, in the cases where the source signals are non-stationary. Real situations often involve difficult non-stationarity such that each source signal appears or disappears and hence the number of sources varies with time. In the present study, we address such "source switching problems" to construct a robust ICA algorithm that deals with non-stationary signals. We formulate a generative model that incorporates source switching and noisy mixing, and then derive a noisy ICA algorithm based on the Bayesian inference. In a simple simulation, it is shown that our method has a potential to resolve the "source switching problems."
Keyword(in Japanese) (See Japanese page)
Keyword(in English) independent component analysis / non-stationary signal / variational Bayes / variable selection
Paper # NC2004-224
Date of Issue

Conference Information
Committee NC
Conference Date 2005/3/23(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 Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Bayesian noisy ICA for source switching environments
Sub Title (in English)
Keyword(1) independent component analysis
Keyword(2) non-stationary signal
Keyword(3) variational Bayes
Keyword(4) variable selection
1st Author's Name Junichiro HIRAYAMA
1st Author's Affiliation Graduate School of Information Science, Nara Institute of Science and Technology()
2nd Author's Name Shinichi MAEDA
2nd Author's Affiliation Graduate School of Information Science, Nara Institute of Science and Technology
3rd Author's Name Shin ISHII
3rd Author's Affiliation Graduate School of Information Science, Nara Institute of Science and Technology
Date 2005/3/23
Paper # NC2004-224
Volume (vol) vol.104
Number (no) 760
Page pp.pp.-
#Pages 5
Date of Issue