Transformer-based LLM4UACP predicts underwater acoustic channels

Researchers have developed LLM4UACP, a novel framework that adapts a pretrained Transformer model for predicting underwater acoustic channels. This dual-domain approach extracts features from both the channel impulse response (CIR) and channel frequency response (CFR), fusing them into delay-time tokens. By using Butterfly Orthogonal Fine-tuning, LLM4UACP can predict multiple future CIRs in a single forward pass, avoiding the error accumulation common in recursive methods. Experiments show significant reductions in normalized mean squared error compared to existing baselines, even with limited data and cross-dataset transfer. AI

IMPACT This research could improve the reliability and efficiency of underwater acoustic communications through advanced prediction techniques.

RANK_REASON The item is an academic paper detailing a new model and methodology for a specific technical problem. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• ISSod3 transposase TnpA_ISSod3 SO_0146
• LLM4UACP
• ScienceCast
• transformer
• transposase OrfA SO_A0025

• paper
• infra

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