Reweighted Expectation Maximization

Abstract

Training deep generative models with maximum likelihood remains a challenge. The typical workaround is to use variational inference (VI) and maximize a lower bound to the log marginal likelihood of the data. Variational auto-encoders (VAEs) adopt this approach. They further amortize the cost of inference by using a recognition network to parameterize the variational family. Amortized VI scales approximate posterior inference in deep generative models to large datasets. However it introduces an amortization gap and leads to approximate posteriors of reduced expressivity due to the problem known as posterior collapse. In this paper, we consider expectation maximization (EM) as a paradigm for fitting deep generative models. Unlike VI, EM directly maximizes the log marginal likelihood of the data. We rediscover the importance weighted auto-encoder (IWAE) as an instance of EM and propose a new EM-based algorithm for fitting deep generative models called reweighted expectation maximization (REM). REM learns better generative models than the IWAE by decoupling the learning dynamics of the generative model and the recognition network using a separate expressive proposal found by moment matching. We compared REM to the VAE and the IWAE on several density estimation benchmarks and found it leads to significantly better performance as measured by log-likelihood.

arXiv commentCode can be found at https://github.com/adjidieng/REM

References in corpus (11)

• Auto-Encoding Variational Bayes
• Stochastic Backpropagation and Approximate Inference in Deep Generative Models
• Variational Inference with Normalizing Flows
• DRAW: A Recurrent Neural Network For Image Generation
• Density estimation using Real NVP
• Variational Lossy Autoencoder
• Lagging Inference Networks and Posterior Collapse in Variational Autoencoders
• Deep AutoRegressive Networks
• Inference Suboptimality in Variational Autoencoders
• Auto-Encoding Sequential Monte Carlo
• Revisiting Reweighted Wake-Sleep for Models with Stochastic Control Flow

Cited by in corpus (7)

• Prescribed Generative Adversarial Networks
• A likelihood approach to nonparametric estimation of a singular distribution using deep generative models
• Markovian Score Climbing: Variational Inference with KL(p||q)
• Unbiased Gradient Estimation for Variational Auto-Encoders using Coupled Markov Chains
• Amortised Learning by Wake-Sleep
• Generative Text Modeling through Short Run Inference
• Analytical Probability Distributions and EM-Learning for Deep Generative Networks