Deepcode: Feedback Codes via Deep Learning

TL;DR

This paper introduces the first deep learning-based codes for Gaussian noise channels with feedback, significantly outperforming traditional codes and demonstrating adaptability and scalability.

Contribution

It presents a novel family of feedback codes created via deep learning that surpass existing codes in reliability and incorporate information-theoretic insights.

Findings

Outperforms state-of-the-art codes by 3 orders of magnitude in reliability

Generalizes to larger block lengths and practical constraints

Can be combined with known codes for enhanced performance

Abstract

The design of codes for communicating reliably over a statistically well defined channel is an important endeavor involving deep mathematical research and wide-ranging practical applications. In this work, we present the first family of codes obtained via deep learning, which significantly beats state-of-the-art codes designed over several decades of research. The communication channel under consideration is the Gaussian noise channel with feedback, whose study was initiated by Shannon; feedback is known theoretically to improve reliability of communication, but no practical codes that do so have ever been successfully constructed. We break this logjam by integrating information theoretic insights harmoniously with recurrent-neural-network based encoders and decoders to create novel codes that outperform known codes by 3 orders of magnitude in reliability. We also demonstrate several desirable properties of the codes: (a) generalization to larger block lengths, (b) composability with known codes, (c) adaptation to practical constraints. This result also has broader ramifications for coding theory: even when the channel has a clear mathematical model, deep learning methodologies, when combined with channel-specific information-theoretic insights, can potentially beat state-of-the-art codes constructed over decades of mathematical research.