Learning aligned EEG representations with subject-specific encoders
For EEG decoding, subject-specific encoders offer a learned alignment mechanism that reduces the need for explicit preprocessing, but the approach is incremental over existing methods.
Subject-specific encoders with a shared classifier learn to align EEG representations across subjects, matching or exceeding Euclidean Alignment on motor-imagery tasks, improving most subjects while leaving a method-sensitive subset as a bottleneck.
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.