Unified Signal Compression Using Generative Adversarial Networks

TL;DR

This paper introduces a unified GAN-based framework for compressing image and speech signals, achieving superior quality and efficiency compared to previous methods through innovative quantization and optimization techniques.

Contribution

It presents a novel unified compression approach using GANs with optimized latent vector quantization, improving performance across multiple signal types.

Findings

Outperforms prior methods in bit rate and PSNR

Achieves higher neural network classification accuracy

Effective for both image and speech compression

Abstract

We propose a unified compression framework that uses generative adversarial networks (GAN) to compress image and speech signals. The compressed signal is represented by a latent vector fed into a generator network which is trained to produce high quality signals that minimize a target objective function. To efficiently quantize the compressed signal, non-uniformly quantized optimal latent vectors are identified by iterative back-propagation with ADMM optimization performed for each iteration. Our experiments show that the proposed algorithm outperforms prior signal compression methods for both image and speech compression quantified in various metrics including bit rate, PSNR, and neural network based signal classification accuracy.