Revisiting Reweighted Wake-Sleep for Models with Stochastic Control Flow

TL;DR

This paper demonstrates that the reweighted wake-sleep algorithm outperforms current methods in training stochastic control-flow models, especially as the number of particles increases, making it a strong alternative for such models.

Contribution

The paper revisits and extensively evaluates the reweighted wake-sleep algorithm, showing its superiority over state-of-the-art methods for learning stochastic control-flow models.

Findings

RWS outperforms current methods in learning SCFMs.

RWS learns better models with increasing particles.

RWS is a competitive alternative for training SCFMs.

Abstract

Stochastic control-flow models (SCFMs) are a class of generative models that involve branching on choices from discrete random variables. Amortized gradient-based learning of SCFMs is challenging as most approaches targeting discrete variables rely on their continuous relaxations---which can be intractable in SCFMs, as branching on relaxations requires evaluating all (exponentially many) branching paths. Tractable alternatives mainly combine REINFORCE with complex control-variate schemes to improve the variance of naive estimators. Here, we revisit the reweighted wake-sleep (RWS) (Bornschein and Bengio, 2015) algorithm, and through extensive evaluations, show that it outperforms current state-of-the-art methods in learning SCFMs. Further, in contrast to the importance weighted autoencoder, we observe that RWS learns better models and inference networks with increasing numbers of particles. Our results suggest that RWS is a competitive, often preferable, alternative for learning SCFMs.