PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

An Algebraic Method for Approximate Rank One Factorization of Rank Deficient Matrices
Franz Kiraly, Andreas Ziehe and Klaus-Robert Müller
In: LVA ICA 2012, 12-15 Mar 2012, Tel-Aviv, Israel.

Abstract

In this paper we consider the problem of finding approximate common rank one factors for a set of matrices. Instead of jointly diagonalizing the matrices, we perform calculations directly in the problem intrinsic domain: we present an algorithm, AROFAC, which searches the approximate linear span of the matrices using an indicator function for the rank one factors, finding specific single sources. We evaluate the feasibility of this approach by discussing simulations on generated data and a neurophysiological dataset. Note however that our contribution is intended to be mainly conceptual in nature.

EPrint Type:Conference or Workshop Item (Paper)
Additional Information:@incollection {KirZieMue12, author = {Király, Franz and Ziehe, Andreas and Müller, Klaus-Robert}, affiliation = {Machine Learning Group, Technische Universität Berlin, Franklinstr. 28/29, 10587 Berlin, Germany}, title = {An Algebraic Method for Approximate Rank One Factorization of Rank Deficient Matrices}, booktitle = {Latent Variable Analysis and Signal Separation}, series = {Lecture Notes in Computer Science}, editor = {Theis, Fabian and Cichocki, Andrzej and Yeredor, Arie and Zibulevsky, Michael}, publisher = {Springer Berlin / Heidelberg}, isbn = {978-3-642-28550-9}, pages = {272-279}, volume = {7191}, url = {http://dx.doi.org/10.1007/978-3-642-28551-6_34}, year = {2012} }
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Learning/Statistics & Optimisation
Theory & Algorithms
ID Code:9430
Deposited By:Klaus-Robert Müller
Deposited On:16 March 2012