New In-Context Decoding Framework Enhances Text Transmission Robustness

Researchers have developed a novel In-Context Decoding (ICD) framework designed to enhance the robustness of Separate Source-Channel Coding (SSCC) for text transmission. This receiver-side approach utilizes an Error Correction Code Transformer (ECCT) to assess the reliability of decoded bits. The system then generates a pool of candidate reconstructions by performing reliability-guided bit flipping, samples from this pool, and employs a large-language model-based arithmetic decoder to produce final outputs and their associated log-likelihoods. Experiments show that ICD consistently outperforms conventional SSCC methods and some Joint Source-Channel Coding (JSCC) schemes across various challenging channel conditions. AI

IMPACT This research could lead to more reliable text transmission in challenging communication environments by leveraging LLMs for error correction.

RANK_REASON The cluster contains a research paper detailing a new technical framework for improving data transmission. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• additive white Gaussian noise
• arithmetic coding
• Error Correction Code Transformer
• In-Context Decoding
• In-Context Source and Channel Coding
• large-language models
• Separate Source-Channel Coding
• Ziqiong Wang

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →