Incorporating latent or hidden variables is a crucial aspect of statistical modeling. I will describe wide applicability of tensor decomposition techniques for learning many popular models including topic models, hidden Markov models, network community models and Gaussian mixtures. In addition to possessing strong theoretical guarantees, the tensor methods are fast, accurate and highly parallelizable in practice.
Session Summary
Incorporating latent or hidden variables is a crucial aspect of statistical modeling
MLconf 2014 New York City
Anima Anandkumar
Bren professor at Caltech CMS department and a director of machine learning research at NVIDIA.
Learn more »