Three scatter matrices and independent subspace analysis

Abstract

In independent subspace analysis (ISA) one assumes that the components of the observed random vector are linear combinations of the components of a latent random vector with independent subvectors. The problem is then to Їnd an estimate of a transformation matrix to recover the independent subvectors. Regular independent component analysis (ICA) is a special case. In this paper we show how three scatter matrices with the so called block independence property can be used in independent subspace analysis. The procedure is illustrated with a small example.