April 8, 2014
Anirban Bhattacharya
Assistant Professor of Statistics, Texas A&M University
Title: Tensor decompositions and sparse log-linear models.
Abstract:
Contingency table analysis routinely relies on log linear models, with latent structure analysis providing a common alternative. Latent structure models lead to a low rank tensor factorization of the probability mass function for multivariate categorical data, while log linear models achieve dimensionality reduction through sparsity. Little is known about the relationship between these notions of dimensionality reduction in the two paradigms. In particular, we are interested to know whether sparse log-linear models can be expressed parsimoniously via a latent structure model. We discuss some initial results in this direction.