Article ID Journal Published Year Pages File Type
425758 Future Generation Computer Systems 2006 11 Pages PDF
Abstract

Fixed-point algorithms for neural-network learning have gained increasing interest over recent years. In blind signal processing by neural systems, they normally appear in the literature as one-unit learning rules. The aim of the present short contribution is to introduce some fully-parallel (multi-unit) learning algorithms on the group of orthogonal matrices to be applied to independent component analysis.

Related Topics
Physical Sciences and Engineering Computer Science Computational Theory and Mathematics
Authors
,