Learning Discrete Structured Variational Auto-Encoder using Natural Evolution Strategies
This work addresses a computational bottleneck in generative learning for researchers and practitioners dealing with discrete structured VAEs, but it is incremental as it applies an existing optimization method to a known problem.
The paper tackled the problem of learning discrete structured variational auto-encoders (VAEs) by using Natural Evolution Strategies (NES) to avoid gradient propagation through high-dimensional discrete latent spaces, and demonstrated empirically that NES is as effective as gradient-based approximations while proving convergence for non-Lipschitz functions.
Discrete variational auto-encoders (VAEs) are able to represent semantic latent spaces in generative learning. In many real-life settings, the discrete latent space consists of high-dimensional structures, and propagating gradients through the relevant structures often requires enumerating over an exponentially large latent space. Recently, various approaches were devised to propagate approximated gradients without enumerating over the space of possible structures. In this work, we use Natural Evolution Strategies (NES), a class of gradient-free black-box optimization algorithms, to learn discrete structured VAEs. The NES algorithms are computationally appealing as they estimate gradients with forward pass evaluations only, thus they do not require to propagate gradients through their discrete structures. We demonstrate empirically that optimizing discrete structured VAEs using NES is as effective as gradient-based approximations. Lastly, we prove NES converges for non-Lipschitz functions as appear in discrete structured VAEs.