Saved in:
Bibliographic Details
Main Authors: Lau, Chin Wa, Shi, Xiang, Zheng, Ziyan, Cao, Haiwen, Guo, Nian
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.15637
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911163404517376
author Lau, Chin Wa
Shi, Xiang
Zheng, Ziyan
Cao, Haiwen
Guo, Nian
author_facet Lau, Chin Wa
Shi, Xiang
Zheng, Ziyan
Cao, Haiwen
Guo, Nian
contents Transformer-based neural decoders have emerged as a promising approach to error correction coding, combining data-driven adaptability with efficient modeling of long-range dependencies. This paper presents a novel decoder architecture that integrates classical belief propagation principles with transformer designs. We introduce a differentiable syndrome loss function leveraging global codebook structure and a differential-attention mechanism optimizing bit and syndrome embedding interactions. Experimental results demonstrate consistent performance improvements over existing transformer-based decoders, with our approach surpassing traditional belief propagation decoders for short-to-medium length LDPC codes.
format Preprint
id arxiv_https___arxiv_org_abs_2509_15637
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Interplay Between Belief Propagation and Transformer: Differential-Attention Message Passing Transformer
Lau, Chin Wa
Shi, Xiang
Zheng, Ziyan
Cao, Haiwen
Guo, Nian
Information Theory
Signal Processing
Transformer-based neural decoders have emerged as a promising approach to error correction coding, combining data-driven adaptability with efficient modeling of long-range dependencies. This paper presents a novel decoder architecture that integrates classical belief propagation principles with transformer designs. We introduce a differentiable syndrome loss function leveraging global codebook structure and a differential-attention mechanism optimizing bit and syndrome embedding interactions. Experimental results demonstrate consistent performance improvements over existing transformer-based decoders, with our approach surpassing traditional belief propagation decoders for short-to-medium length LDPC codes.
title Interplay Between Belief Propagation and Transformer: Differential-Attention Message Passing Transformer
topic Information Theory
Signal Processing
url https://arxiv.org/abs/2509.15637