DVAE++: Discrete Variational Autoencoders with Overlapping Transformations
Training of discrete latent variable models remains challenging because passing gradient information through discrete units is difficult. We propose a new class of smoothing transformations based on a mixture of two overlapping distributions, and show that the proposed transformation can be used for training binary latent models with either directed or undirected priors. We derive a new variational bound to efficiently train with Boltzmann machine priors. Using this bound, we develop DVAE++, a generative model with a global discrete prior and a hierarchy of convolutional continuous variables. Experiments on several benchmarks show that overlapping transformations outperform other recent continuous relaxations of discrete latent variables including Gumbel-Softmax (Maddison et al., 2016; Jang et al., 2016), and discrete variational autoencoders (Rolfe 2016).
PDF Abstract ICML 2018 PDF ICML 2018 AbstractTasks
Datasets
CIFAR-10
MNIST
Silhouettes
Results from the Paper
Ranked #53 on
Image Generation
on CIFAR-10
(bits/dimension metric)
