Stick-Breaking Variational Autoencoders

Code

• github.com/enalisnick/stick-breaking_dgms
  OfficialIn papernone★ 96
• github.com/kyeongry/Autoencoders
  pytorch★ 0

Abstract

We extend Stochastic Gradient Variational Bayes to perform posterior inference for the weights of Stick-Breaking processes. This development allows us to define a Stick-Breaking Variational Autoencoder (SB-VAE), a Bayesian nonparametric version of the variational autoencoder that has a latent representation with stochastic dimensionality. We experimentally demonstrate that the SB-VAE, and a semi-supervised variant, learn highly discriminative latent representations that often outperform the Gaussian VAE's.

Reproductions