PTSM: Physiology-aware and Task-invariant Spatio-temporal Modeling for Cross-Subject EEG Decoding

2025-08-15

AI Summary

Cross-subject EEG decoding faces two core challenges: high inter-subject variability and scarcity of shared, task-invariant neural representations. To address these, we propose an interpretable and robust end-to-end decoding framework featuring a novel dual-branch masking mechanism—separately modeling subject-specific and task-invariant spatiotemporal dynamics—and enforcing orthogonal disentanglement of these factors in the latent space via information-theoretic regularization. Our method integrates spatiotemporal decomposition masks, multi-objective loss optimization, contrastive learning, and mutual information regularization to enable zero-shot cross-subject transfer without subject-specific calibration. Evaluated on multiple motor imagery datasets, it consistently outperforms state-of-the-art methods in accuracy and generalizability. The framework establishes a new paradigm for universal brain–computer interfaces by enabling robust, calibration-free decoding across diverse subjects.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Multimodal Learning
• Computer Vision: Interpretability, Explainability, and Transparency

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Cross-subject electroencephalography (EEG) decoding remains a fundamental challenge in brain-computer interface (BCI) research due to substantial inter-subject variability and the scarcity of subject-invariant representations. This paper proposed PTSM (Physiology-aware and Task-invariant Spatio-temporal Modeling), a novel framework for interpretable and robust EEG decoding across unseen subjects. PTSM employs a dual-branch masking mechanism that independently learns personalized and shared spatio-temporal patterns, enabling the model to preserve individual-specific neural characteristics while extracting task-relevant, population-shared features. The masks are factorized across temporal and spatial dimensions, allowing fine-grained modulation of dynamic EEG patterns with low computational overhead. To further address representational entanglement, PTSM enforces information-theoretic constraints that decompose latent embeddings into orthogonal task-related and subject-related subspaces. The model is trained end-to-end via a multi-objective loss integrating classification, contrastive, and disentanglement objectives. Extensive experiments on cross-subject motor imagery datasets demonstrate that PTSM achieves strong zero-shot generalization, outperforming state-of-the-art baselines without subject-specific calibration. Results highlight the efficacy of disentangled neural representations for achieving both personalized and transferable decoding in non-stationary neurophysiological settings.

Problem

• Addresses cross-subject EEG decoding challenges due to variability
• Proposes PTSM for interpretable, robust EEG decoding across subjects
• Enhances generalization via disentangled neural representations in BCIs

Innovation

• Dual-branch masking for spatio-temporal EEG patterns
• Factorization of masks across time and space
• Information-theoretic constraints for disentangled embeddings