X-REFINE: XAI-based RElevance input-Filtering and archItecture fiNe-tuning for channel Estimation
Researchers have developed X-REFINE, a novel framework that enhances the interpretability and efficiency of AI models in wireless communications. This XAI-based approach jointly filters irrelevant inputs and fine-tunes the model architecture by analyzing the relevance of subcarriers and hidden neurons. X-REFINE demonstrates a superior trade-off between performance, complexity, and interpretability compared to existing methods, significantly reducing computational load while maintaining strong bit error rate performance. AI
IMPACT This framework could lead to more efficient and interpretable AI models for critical applications like 6G wireless communications.