Learning sparse auto-encoders for green AI image coding

TL;DR

This paper introduces a structured sparse learning approach for convolutional auto-encoders in image coding, achieving comparable performance to dense networks while significantly reducing memory and computational costs.

Contribution

It proposes a new structured sparse learning method with an $ ext{l}_{1,1}$ constraint for efficient image coding auto-encoders, improving on prior dense models.

Findings

The $ ext{l}_{1,1}$ constraint yields the best structured sparsity.

The method reduces memory and computational costs significantly.

Performance remains comparable to dense networks in rate-distortion.

Abstract

Recently, convolutional auto-encoders (CAE) were introduced for image coding. They achieved performance improvements over the state-of-the-art JPEG2000 method. However, these performances were obtained using massive CAEs featuring a large number of parameters and whose training required heavy computational power.\\ In this paper, we address the problem of lossy image compression using a CAE with a small memory footprint and low computational power usage. In order to overcome the computational cost issue, the majority of the literature uses Lagrangian proximal regularization methods, which are time consuming themselves.\\ In this work, we propose a constrained approach and a new structured sparse learning method. We design an algorithm and test it on three constraints: the classical $\ell_1$ constraint, the $\ell_{1,\infty}$ and the new $\ell_{1,1}$ constraint. Experimental results show that the $\ell_{1,1}$ constraint provides the best structured sparsity, resulting in a high reduction of memory and computational cost, with similar rate-distortion performance as with dense networks.