MPNet: A Robust and Efficient Manifold Pooling Network for Multi-Rhythm EEG Signal Decoding

Collection of recent arXiv EESS.SP papers (May 8, 2026) covering advances in signal processing and applied ML for biomedical sensing, wireless communications, and estimation. Highlights include: MPNet, a manifold-pooling Riemannian network for multi-rhythm EEG that reduces Riemannian input dimensionality and speeds inference up to 10x; MedMamba, a multi-scale bidirectional state-space model outperforming Transformers on ECG/EEG and enabling 4.6x inference speedups; a PPO-based DRL framework for dynamic HAPS base-station positioning robust to stratospheric wind; a KD-based, memory-efficient EDA