The authors say MVCLDG fuses raw electroencephalography with phase information derived from the Hilbert transform via multi-scale inception blocks, capturing amplitude and phase features, then applies domain-alignment and contrastive-learning constraints to reduce cross-domain distributional discrepancy, compact within-class representations and enlarge between-class separability. Evaluated on a public Error-Related Negativity (ERN) dataset and a self-collected semantic-syntactic violation dataset, it outperformed baseline and representative domain generalization methods in cross-subject settings; Eigen-Class Activation Maps visualizations showed consistency between model-attended electrodes and known neurophysiological scalp patterns, supporting its biological interpretability.
Multi-View Contrastive Learning Improves Cross-Subject ERP Classification
Summary
MVCLDG combines raw EEG and Hilbert-derived phase information with domain-alignment and contrastive-learning constraints to improve classification across unseen users. It outperformed baseline and representative domain-generalization methods on a public error-related-negativity dataset and a semantic-syntactic-violation dataset without target-domain adaptation; ablation and activation-map analyses supported the contribution and neurophysiological plausibility of its components.
Why it matters
Cross-subject generalization is a core obstacle for ERP-based closed-loop BCIs, and the no-adaptation requirement makes this phase-aware approach practically relevant.
Compiled by BCIwiki from public sources
Sources · 1
pubmed.ncbi.nlm.nih.gov 2026-06-27