A Learnable SIM Paradigm: Fundamentals, Training Techniques, and Applications

This collection of recent papers focuses on advanced physical-layer and edge ML technologies for next-generation wireless systems: learnable stacked intelligent metasurfaces (SIMs) for in-hardware analog computing and multi-user signal separation/anti‑jamming; analog microwave computing implementations (hybrid couplers + phase shifters) for very-low-latency transforms; BD‑RIS joint training and other RIS/beamforming advances that drastically reduce pilot overhead; near-field beam training robust to multipath via hybrid learning+optimization; federated/ hybrid Transformer approaches for human-­