Unscented Autoencoder

TL;DR

The paper introduces the Unscented Autoencoder, a novel deep generative model that uses the Unscented Transform for improved posterior representation and Wasserstein metric for sharper reconstructions, demonstrating competitive results.

Contribution

It proposes a new deterministic sampling approach for VAEs, called the Unscented Autoencoder, combining the Unscented Transform and Wasserstein metric for better posterior estimation and training stability.

Findings

Achieves competitive FID scores compared to related models.

Exhibits lower training variance than traditional VAEs.

Utilizes deterministic sigma points for improved posterior approximation.

Abstract

The Variational Autoencoder (VAE) is a seminal approach in deep generative modeling with latent variables. Interpreting its reconstruction process as a nonlinear transformation of samples from the latent posterior distribution, we apply the Unscented Transform (UT) -- a well-known distribution approximation used in the Unscented Kalman Filter (UKF) from the field of filtering. A finite set of statistics called sigma points, sampled deterministically, provides a more informative and lower-variance posterior representation than the ubiquitous noise-scaling of the reparameterization trick, while ensuring higher-quality reconstruction. We further boost the performance by replacing the Kullback-Leibler (KL) divergence with the Wasserstein distribution metric that allows for a sharper posterior. Inspired by the two components, we derive a novel, deterministic-sampling flavor of the VAE, the Unscented Autoencoder (UAE), trained purely with regularization-like terms on the per-sample posterior. We empirically show competitive performance in Fr\'echet Inception Distance (FID) scores over closely-related models, in addition to a lower training variance than the VAE.