Split Hierarchical Variational Compression

TL;DR

This paper introduces Split Hierarchical Variational Compression (SHVC), a novel VAE-based method that improves image compression efficiency and performance by enabling parallel coding and reducing model complexity.

Contribution

SHVC proposes an autoregressive prior and a flexible coding framework that overcome practical limitations of existing VAE-based compression methods.

Findings

Achieves state-of-the-art lossless image compression performance.

Uses up to 100x fewer model parameters than competing VAEs.

Supports parallel coding, enhancing practical deployment.

Abstract

Variational autoencoders (VAEs) have witnessed great success in performing the compression of image datasets. This success, made possible by the bits-back coding framework, has produced competitive compression performance across many benchmarks. However, despite this, VAE architectures are currently limited by a combination of coding practicalities and compression ratios. That is, not only do state-of-the-art methods, such as normalizing flows, often demonstrate out-performance, but the initial bits required in coding makes single and parallel image compression challenging. To remedy this, we introduce Split Hierarchical Variational Compression (SHVC). SHVC introduces two novelties. Firstly, we propose an efficient autoregressive prior, the autoregressive sub-pixel convolution, that allows a generalisation between per-pixel autoregressions and fully factorised probability models. Secondly, we define our coding framework, the autoregressive initial bits, that flexibly supports parallel coding and avoids -- for the first time -- many of the practicalities commonly associated with bits-back coding. In our experiments, we demonstrate SHVC is able to achieve state-of-the-art compression performance across full-resolution lossless image compression tasks, with up to 100x fewer model parameters than competing VAE approaches.