Structure Learning by Pruning in Independent Component Analysis

Andreas Brinch Nielsen, Lars Kai Hansen

AbstractWe discuss pruning as a means of structure learning in independent component analysis.
Sparse models are attractive in both signal processing and in analysis of abstract data, they
can assist model interpretation, generalizability and reduce computation. We derive the
relevant saliency expressions and compare with magnitude based pruning and Bayesian
sparsification. We show in simulations that pruning is able to identify underlying sparse
structures without prior knowledge on the degree of sparsity. We find that for ICA magnitude
based pruning is as efficient as saliency based methods and Bayesian methods, for
both small and large samples. The Bayesian information criterion (BIC) seems to outperform
both AIC and test sets as tools for determining the optimal degree of sparsity.
KeywordsIndependent component analysis, pruning, sparsity,
TypeTechnical report
Year2006    Month September
BibTeX data [bibtex]
IMM Group(s)Intelligent Signal Processing


Back  ::  IMM Publications