Interpretable EEG Microstate Discovery via Variational Deep Embedding

Presented at XAI 2026 in Fortaleza, Brazil (1-3 July 2026), co-located with the 4th World Conference on eXplainable Artificial Intelligence. The talk covered the Convolutional Variational Deep Embedding (Conv-VaDE) approach and the systematic architecture search across cluster counts, latent dimensionality, network depth, and channel width on the LEMON resting-state EEG dataset. Companion repo with the full sweep code: microstate-architecture-search. See the publication for the full abstract and arXiv link.