DARE-EEG: A Foundation Model for Mining Dual-Aligned Representation of EEG

In three linesDARE-EEG is a self-supervised foundation model for EEG that learns representations invariant to incomplete observations through dual-aligned learning (mask alignment + anchor alignment). Evaluated across multiple EEG benchmarks, it achieves state-of-the-art accuracy with low parameter complexity and superior cross-dataset portability.

Summary generated by Claude — human-verified