Researchers have developed SemQAM, a novel scheme for joint semantic-channel coding and modulation designed to improve generative image transmission. Unlike traditional methods that optimize these components separately, SemQAM integrates them to allow source representations to adapt better to channel impairments. The system uses vector quantization and maps indices to QAM symbols, updating codewords to directly address channel effects and reduce error propagation in autoregressive coding. Experiments indicate SemQAM-C outperforms existing methods like OneDC in image quality metrics while maintaining competitive source coding capabilities, and SemQAM itself reduces channel uses by over 18% compared to separated approaches. AI
IMPACT Improves efficiency and quality in transmitting images generated by AI models.
RANK_REASON The cluster contains two arXiv preprints detailing a new technical method for image transmission.
Read on Hugging Face Daily Papers →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
