Mitigating Subject Dependency in EEG Decoding with Subject-Specific Low-Rank Adapters
This addresses the obstacle of subject dependency for developing foundation models in EEG decoding, offering a practical and scalable solution.
The paper tackles the problem of subject-specific distribution shifts in EEG decoding by proposing Subject-Conditioned Layer, which decomposes weights into shared and subject-specific low-rank components, resulting in models that outperform subject-agnostic and individually trained subject-specific models.
Subject-specific distribution shifts represent an important obstacle to the development of foundation models for EEG decoding. To address this, we propose Subject-Conditioned Layer,, an adaptive layer designed as a drop-in replacement for standard linear or convolutional layers in any neural network architecture. Our layer captures subject-specific variability by decomposing its weights into a shared, subject-invariant component and a lightweight, low-rank correction unique to each subject. This explicit separation of general knowledge from personalized adaptation allows existing models to become robust to subject shifts. Empirically, models equipped with our layer outperform both a shared-weight-only model (subject-agnostic model) and the average of individually trained subject-specific models. Consequently, the Subject-Conditioned Layer, offers a practical and scalable path towards building effective cross-subject foundation models for EEG.