Semantic Communication via Rate Distortion Perception Bottleneck

TL;DR

This paper introduces a semantic communication framework that jointly optimizes perception and bit-level distortion using an information bottleneck approach, addressing perceptual quality issues in data reconstruction.

Contribution

It develops a novel IB-based image inference and reconstruction network that considers semantic-level distortion, expanding traditional Shannon theory to include perceptual quality.

Findings

Confirmed the rate distortion perception tradeoff experimentally on MNIST.

Demonstrated improved perceptual quality in image reconstruction.

Unified optimization of perception and compression achieved.

Abstract

With the advancement of Artificial Intelligence (AI) technology, next-generation wireless communication network is facing unprecedented challenge. Semantic communication has become a novel solution to address such challenges, with enhancing the efficiency of bandwidth utilization by transmitting meaningful information and filtering out superfluous data. Unfortunately, recent studies have shown that classical Shannon information theory primarily focuses on the bit-level distortion, which cannot adequately address the perceptual quality issues of data reconstruction at the receiver end. In this work, we consider the impact of semantic-level distortion on semantic communication. We develop an image inference network based on the Information Bottleneck (IB) framework and concurrently establish an image reconstruction network. This network is designed to achieve joint optimization of perception and bit-level distortion, as well as image inference, associated with compressing information. To maintain consistency with the principles of IB for handling high-dimensional data, we employ variational approximation methods to simplify the optimization problem. Finally, we confirm the existence of the rate distortion perception tradeoff within IB framework through experimental analysis conducted on the MNIST dataset.