ASPEN: Spectral-Temporal Fusion for Cross-Subject Brain Decoding

TL;DR

This paper introduces ASPEN, a hybrid spectral-temporal model that improves cross-subject EEG decoding by leveraging spectral features' stability and a multiplicative fusion approach, achieving superior generalization across multiple datasets.

Contribution

The paper proposes ASPEN, a novel hybrid architecture that combines spectral and temporal EEG features through multiplicative fusion for enhanced cross-subject brain decoding.

Findings

Spectral features show higher cross-subject similarity than temporal signals.

ASPEN achieves state-of-the-art accuracy on three datasets.

Multiplicative fusion enables adaptive spectral-temporal balance.

Abstract

Cross-subject generalization in EEG-based brain-computer interfaces (BCIs) remains challenging due to individual variability in neural signals. We investigate whether spectral representations offer more stable features for cross-subject transfer than temporal waveforms. Through correlation analyses across three EEG paradigms (SSVEP, P300, and Motor Imagery), we find that spectral features exhibit consistently higher cross-subject similarity than temporal signals. Motivated by this observation, we introduce ASPEN, a hybrid architecture that combines spectral and temporal feature streams via multiplicative fusion, requiring cross-modal agreement for features to propagate. Experiments across six benchmark datasets reveal that ASPEN is able to dynamically achieve the optimal spectral-temporal balance depending on the paradigm. ASPEN achieves the best unseen-subject accuracy on three of six datasets and competitive performance on others, demonstrating that multiplicative multimodal fusion enables effective cross-subject generalization.