Presentation 1999/3/19
Independent Component Analysis by Geometric Optimization method
Yasunori Nishimori, Shotaro Akaho,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Various problems arizing from signal processing, computer vision, and control theory can be regarded as opitimization problems on the orthogonal group. But when we apply ordinary iterative optimization methods such as the gradient descent method to those problems, the updated matrix deviates from the orthogonal group in each step. In order to circumvent this problem we propose a new geometric optimization method for the orthogonal group based on geodesics, and applied it to independent component analysis, which is a new statistical signal processing method making impact on a variety of fields. Simulation result shows our method accelerates the covergence of learning.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) ICA / gradient descent method / Newton method / Lie group / geodesics
Paper # NC98-161
Date of Issue

Conference Information
Committee NC
Conference Date 1999/3/19(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) Independent Component Analysis by Geometric Optimization method
Sub Title (in English)
Keyword(1) ICA
Keyword(2) gradient descent method
Keyword(3) Newton method
Keyword(4) Lie group
Keyword(5) geodesics
1st Author's Name Yasunori Nishimori
1st Author's Affiliation Electrotechnical Laboratory()
2nd Author's Name Shotaro Akaho
2nd Author's Affiliation Electrotechnical Laboratory
Date 1999/3/19
Paper # NC98-161
Volume (vol) vol.98
Number (no) 674
Page pp.pp.-
#Pages 7
Date of Issue