AFPM: Alignment-based Frame Patch Modeling for Cross-Dataset EEG Decoding

2025-07-16

AI Summary

Weak cross-dataset generalization of EEG decoding models in brain–computer interfaces (BCIs) stems from inconsistent electrode montages, signal non-stationarity, and insufficient incorporation of neurophysiological priors. To address this, we propose a calibration-free frame-block modeling framework. Our approach introduces an anatomy-informed spatial alignment mechanism—grounded in regional brain connectivity—to harmonize heterogeneous EEG channel layouts across datasets. Subsequently, a spatiotemporal block encoder maps raw EEG signals into standardized spatiotemporal patch representations. Critically, the framework operates end-to-end without subject- or device-specific calibration. Evaluated on motor imagery and event-related potential tasks, it outperforms 17 state-of-the-art methods, achieving absolute accuracy gains of up to 4.40% and 3.58%, respectively. This substantially enhances the robustness and practical deployability of BCIs in real-world settings.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Calibration & Uncertainty Quantification
• Humans and AI: Brain-Sensing and Analysis

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Electroencephalogram (EEG) decoding models for brain-computer interfaces (BCIs) struggle with cross-dataset learning and generalization due to channel layout inconsistencies, non-stationary signal distributions, and limited neurophysiological prior integration. To address these issues, we propose a plug-and-play Alignment-Based Frame-Patch Modeling (AFPM) framework, which has two main components: 1) Spatial Alignment, which selects task-relevant channels based on brain-region priors, aligns EEG distributions across domains, and remaps the selected channels to a unified layout; and, 2) Frame-Patch Encoding, which models multi-dataset signals into unified spatiotemporal patches for EEG decoding. Compared to 17 state-of-the-art approaches that need dataset-specific tuning, the proposed calibration-free AFPM achieves performance gains of up to 4.40% on motor imagery and 3.58% on event-related potential tasks. To our knowledge, this is the first calibration-free cross-dataset EEG decoding framework, substantially enhancing the practicalness of BCIs in real-world applications.

Problem

• Addresses cross-dataset EEG decoding challenges in BCIs
• Resolves channel layout inconsistencies and signal non-stationarity
• Enhances generalization without dataset-specific calibration

Innovation

• Spatial Alignment for EEG channel selection
• Frame-Patch Encoding for unified EEG modeling
• Calibration-free cross-dataset EEG decoding