Leveraging Code Automorphisms for Improved Syndrome-Based Neural Decoding

2026-05-05

AI Summary

This work addresses the limited performance and poor generalization of syndrome-based neural decoding (SBND) in high-rate, short-block-length scenarios. To overcome these limitations, the paper introduces code automorphisms for the first time as a systematic data augmentation strategy during both training and inference. This approach substantially enhances the model’s learning capacity, enabling it to closely approach maximum-likelihood decoding performance even with limited training data. The results reveal that prior studies significantly underestimated the error-correction potential of SBND due to insufficient training. Experimental evaluations demonstrate the effectiveness and superiority of the proposed method in soft-decision decoding, offering a novel perspective for the design of deep learning-based decoders.

Abstract

Syndrome-based neural decoding (SBND) has emerged as a promising deep learning approach for soft-decision decoding of high-rate, short-length codes. However, this approach still has substantial room for improvement. In this paper, we show how to leverage code automorphisms to enhance the ability of existing SBND models to learn and generalize through data augmentation during training and inference. As a result, for the short high-rate codes considered, we obtain models that closely approach MLD performance using small datasets and proper training. Our findings also suggest that many prior results for SBND models in the literature underestimate their true correction capability due to undertraining. Code to reproduce all results is available at: https://github.com/lebidan/sbnd.

Problem

• syndrome-based neural decoding
• code automorphisms
• soft-decision decoding
• short-length codes
• high-rate codes

Innovation

• code automorphisms
• syndrome-based neural decoding
• data augmentation
• maximum likelihood decoding
• short high-rate codes