A Data Transmission Method for Feature Extraction and Semantic Enhancement of Scarce Data

This letter addresses the challenges of enhancing the performance of semantic communication in few-shot learning (FSL) scenarios, a critical need in 6G communication where low-latency and high-throughput data transmission is required. Traditional methods struggle with limited data, as they rely on large datasets for encoding and decoding information. To solve problem, this letter proposes a novel MS-DeepSC framework based on meta-learning with stochastic gradient descent (Meta-SGD), which includes a lightweight S-Receiver receiver inspired by MobileNetV2 and a self-attention dynamic embedding encoding (SA-DEE) technique. These innovations reduce the dependence on large datasets and enhance adaptability to channel noise. The experimental results demonstrate that the proposed approach significantly improves the model’s ability to perform downstream tasks under data-scarce conditions, making it suitable for real-world applications.