Storchastic: A Framework for General Stochastic Automatic Differentiation
This work addresses a bottleneck in reinforcement learning and variational inference by providing a general framework for stochastic AD, though it is incremental as it builds on existing AD techniques.
The authors tackled the limitations of existing stochastic automatic differentiation methods, which were either restricted to continuous variables or high-variance estimators, by introducing Storchastic, a framework that supports diverse gradient estimation methods and reduces variance, provably unbiased for any-order gradients and implemented as a PyTorch library.
Modelers use automatic differentiation (AD) of computation graphs to implement complex Deep Learning models without defining gradient computations. Stochastic AD extends AD to stochastic computation graphs with sampling steps, which arise when modelers handle the intractable expectations common in Reinforcement Learning and Variational Inference. However, current methods for stochastic AD are limited: They are either only applicable to continuous random variables and differentiable functions, or can only use simple but high variance score-function estimators. To overcome these limitations, we introduce Storchastic, a new framework for AD of stochastic computation graphs. Storchastic allows the modeler to choose from a wide variety of gradient estimation methods at each sampling step, to optimally reduce the variance of the gradient estimates. Furthermore, Storchastic is provably unbiased for estimation of any-order gradients, and generalizes variance reduction techniques to higher-order gradient estimates. Finally, we implement Storchastic as a PyTorch library at https://github.com/HEmile/storchastic.