Abstract:

In this paper we present a model which can decompose a probability densities or count data into a set of shift invariant components. We begin by introducing a regular latent variable model and subsequently extend it to deal with shift invariance in order to model more complex inputs. We develop an expectation maximization algorithm for estimating components and present various results on challenging real-world data. We show that this approach is a probabilistic generalization of well known algorithms such as Non-Negative Matrix Factorization and multi-way decompositions, and discuss its advantages over such approaches.