Factorized Variational Autoencoders for Modeling Audience Reactions to Movies

In this paper, we study non-linear tensor factorization methods based on deep variational autoencoders.