Cross-Subject and Cross-Montage EEG Transfer Learning via Individual Tangent Space Alignment and Spatial-Riemannian Feature Fusion

2025-08-11

AI Summary

To address poor cross-subject and cross-channel EEG generalizability and lengthy calibration in brain–computer interface (BCI)-based personalized music interventions, this paper proposes Individualized Tangent Space Alignment (ITSA), a geometric method that jointly performs distribution matching and supervised rotational alignment to achieve consistent mapping of inter-subject covariance manifolds. Furthermore, we design a hybrid parallel–serial architecture that jointly regularizes Common Spatial Patterns (CSP) and Riemannian geometric features, balancing discriminability and manifold structure preservation. Under leave-one-subject-out cross-validation, the model significantly improves cross-subject classification accuracy, with the parallel fusion variant achieving optimal performance. The approach demonstrates strong robustness to electrode configuration variations and data scarcity. Source code will be made publicly available.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Learning with Manifolds
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Personalised music-based interventions offer a powerful means of supporting motor rehabilitation by dynamically tailoring auditory stimuli to provide external timekeeping cues, modulate affective states, and stabilise gait patterns. Generalisable Brain-Computer Interfaces (BCIs) thus hold promise for adapting these interventions across individuals. However, inter-subject variability in EEG signals, further compounded by movement-induced artefacts and motor planning differences, hinders the generalisability of BCIs and results in lengthy calibration processes. We propose Individual Tangent Space Alignment (ITSA), a novel pre-alignment strategy incorporating subject-specific recentering, distribution matching, and supervised rotational alignment to enhance cross-subject generalisation. Our hybrid architecture fuses Regularised Common Spatial Patterns (RCSP) with Riemannian geometry in parallel and sequential configurations, improving class separability while maintaining the geometric structure of covariance matrices for robust statistical computation. Using leave-one-subject-out cross-validation, `ITSA' demonstrates significant performance improvements across subjects and conditions. The parallel fusion approach shows the greatest enhancement over its sequential counterpart, with robust performance maintained across varying data conditions and electrode configurations. The code will be made publicly available at the time of publication.

Problem

• Addresses EEG signal variability across subjects and montages
• Improves cross-subject BCI generalization for motor rehabilitation
• Enhances class separability with spatial-Riemannian feature fusion

Innovation

• Individual Tangent Space Alignment for EEG
• Fusion of RCSP and Riemannian geometry
• Parallel spatial-Riemannian feature fusion