Learned transform compression with optimized entropy encoding

TL;DR

This paper introduces a novel learned transform compression method that jointly optimizes the transform and entropy coding using relaxed quantization and probability assignment, leading to improved compression efficiency.

Contribution

It presents a completely new approach to learned transform compression by relaxing quantization and probability optimization for end-to-end training.

Findings

Proof-of-concept experiments demonstrate the effectiveness of the proposed methods.

Joint optimization of transform and entropy coding improves compression performance.

Abstract

We consider the problem of learned transform compression where we learn both, the transform as well as the probability distribution over the discrete codes. We utilize a soft relaxation of the quantization operation to allow for back-propagation of gradients and employ vector (rather than scalar) quantization of the latent codes. Furthermore, we apply similar relaxation in the code probability assignments enabling direct optimization of the code entropy. To the best of our knowledge, this approach is completely novel. We conduct a set of proof-of concept experiments confirming the potency of our approaches.