Learning aligned EEG representations with subject-specific encoders
2026-06-15 • Machine Learning
Machine LearningArtificial Intelligence
AI summaryⓘ
The authors explored a new way to improve EEG brain signal decoding across multiple people by using individual encoders for each person instead of one shared encoder. They found that their method naturally aligns differences between people’s brain signals without needing extra adjustments. This approach helped make the brain activity patterns easier to identify for most subjects, though some challenges remain in handling new, unseen individuals. Overall, the study suggests using personal encoders can help EEG models work better across different people.
EEG decodingcross-subjectneural networksmotor imagerysubject-specific encodersEuclidean alignmentlatent manifoldEEGNetrepresentation learning
Authors
Bruna J. Lopes, Gabriel Schwartz, Sylvain Chevallier, Raphael Y. de Camargo, Bruno Aristimunha
Abstract
Cross-subject EEG decoding promises more training data, but it also exposes neural networks to strong inter-subject distribution shifts. We study whether task supervision and architecture alone can learn subject-aligned representations. We replace a shared EEG encoder with subject-specific encoders followed by a common classifier, and compare this hybrid model with standard EEGNet, AttentionBaseNet, and CTNet baselines with Euclidean Alignment (EA) on four motor-imagery datasets. EA improves shared encoders by recentering subject covariances, but the hybrid encoder largely internalises this role: validation-loss curves and latent-distance analyses change little when EA is removed. Subject-specific heads increase class distinctiveness and place each subject close to its own latent manifold, improving most subjects while leaving a method-sensitive subset. These results support subject-specific encoders as a learned alignment mechanism for EEG decoding and identify head selection for unseen subjects as the remaining bottleneck.