Sparse kernel orthonormalized PLS for feature extraction in large datasets



AbstractIn this paper we are presenting a novel multivariate analysis method for large
scale problems. Our scheme is based on a novel kernel orthonormalized partial
least squares (PLS) variant for feature extraction, imposing sparsity constrains
in the solution to improve scalability. The algorithm is tested on a benchmark
of UCI data sets, and on the analysis of integrated short-time music features for
genre prediction. The upshot is that the method has strong expressive power even
with rather few features, is clearly outperforming the ordinary kernel PLS, and
therefore is an appealing method for feature extraction of labelled data.
TypeConference paper [With referee]
ConferenceNIPS 2006
Year2006    Month December
Electronic version(s)[pdf]
BibTeX data [bibtex]
IMM Group(s)Intelligent Signal Processing