EVA-Net: Interpretable Anomaly Detection for Brain Health via Learning Continuous Aging Prototypes from One-Class EEG Cohorts

We propose EVA-Net to capture multi-scale temporal dependencies in long EEG sequences while leveraging a variational information bottleneck and continuous prototype alignment to learn an interpretable healthy manifold for brain health anomaly detection.

Authors: Kunyu Zhang, Mingxuan Wang, Xiangjie Shi, Haoxing Xu, Chao Zhang

Links: [Paper] [Code]