Variational Rejection Sampling
arXiv:1804.01712
Abstract
Learning latent variable models with stochastic variational inference is challenging when the approximate posterior is far from the true posterior, due to high variance in the gradient estimates. We propose a novel rejection sampling step that discards samples from the variational posterior which are assigned low likelihoods by the model. Our approach provides an arbitrarily accurate approximation of the true posterior at the expense of extra computation. Using a new gradient estimator for the resulting unnormalized proposal distribution, we achieve average improvements of 3.71 nats and 0.21 nats over state-of-the-art single-sample and multi-sample alternatives respectively for estimating marginal log-likelihoods using sigmoid belief networks on the MNIST dataset.
arXiv commentAISTATS 2018
References in corpus (13)
- Auto-Encoding Variational Bayes
- Categorical Reparameterization with Gumbel-Softmax
- Stochastic Backpropagation and Approximate Inference in Deep Generative Models
- The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables
- DRAW: A Recurrent Neural Network For Image Generation
- Markov Chain Monte Carlo and Variational Inference: Bridging the Gap
- Learning in Implicit Generative Models
- REBAR: Low-variance, unbiased gradient estimates for discrete latent variable models
- Backpropagation through the Void: Optimizing control variates for black-box gradient estimation
- Automated Variational Inference in Probabilistic Programming
- Reducing Reparameterization Gradient Variance
- Techniques for Learning Binary Stochastic Feedforward Neural Networks
- Tight Variational Bounds via Random Projections and I-Projections