Deep Bayesian Nonparametric Factor Analysis

TL;DR

This paper introduces a deep Bayesian nonparametric factor analysis model with a beta process prior, enabling the approximation of complex distributions over latent variables, along with a scalable inference algorithm and initial experimental results.

Contribution

It presents a novel deep generative model with a beta process prior and a stochastic EM inference method for complex latent distributions.

Findings

Preliminary results demonstrate the model's potential.

The scalable inference algorithm effectively handles complex models.

The approach advances nonparametric Bayesian deep learning.

Abstract

We propose a deep generative factor analysis model with beta process prior that can approximate complex non-factorial distributions over the latent codes. We outline a stochastic EM algorithm for scalable inference in a specific instantiation of this model and present some preliminary results.