Reintroducing Straight-Through Estimators as Principled Methods for Stochastic Binary Networks

Alexander Shekhovtsov, Viktor Yanush

Keywords: straight-through, binary, stochastic binary, mirror descent

Abstract:

Training neural networks with binary weights and activations is a challenging problem due to the lack of gradients and difficulty of optimization over discrete weights.
Many successful experimental results have been achieved with empirical straight-through (ST) approaches, proposing a variety of ad-hoc rules for propagating gradients through non-differentiable activations and updating discrete weights. At the same time, ST methods can be truly derived as estimators in the stochastic binary network (SBN) model with Bernoulli weights. We advance these derivations to a more complete and systematic study. We analyze properties, estimation accuracy, obtain different forms of correct ST estimators for activations and weights, explain existing empirical approaches and their shortcomings, explain how latent weights arise from the mirror descent method when optimizing over probabilities. This allows to reintroduce, once empirical, ST methods as sound approximations, apply them with clarity and develop further improvements.

One-sentence Summary: Straight-through estimators, wide-spread in the empirical form, are given a proper theoretical treatment for the first time.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Reviewed Version (pdf): https://openreview.net/references/pdf?id=jeKP8Kx31O