Reliable routing improves motor imagery EEG decoding without user labels
EEG-Fusion: Failure-Informed Source-Free Expert Routing for Robust Motor Imagery EEG Decoding
Human-Computer InteractionArtificial IntelligenceMachine Learning
Summary
Motor imagery EEG decoding tries to understand brain signals when a person imagines moving. Sometimes, the decoding fails for certain users even if it looks good on average. The authors created EEG-Fusion, a new system that combines different expert methods and decides which to trust by checking their reliability without knowing the correct answers. This approach improves decoding accuracy and reduces failure for individual users when adapting to new users without labeled data.
What this means in practice
- •For bci developers: Improve motor imagery EEG signal decoding reliability for subject-independent brain-computer interfaces without requiring user-labeled data during adaptation.
- •For rehabilitation device makers: Increase robustness of EEG-based motor imagery control systems in assistive devices for new users without costly calibration sessions.$Commercial implications: This enables commercialization of out-of-the-box brain-controlled assistive technologies that adapt without needing user-specific labels.
Authors
Abdul Basit, Saim Rehman, Muhammad Shafique
Abstract
Subject-independent motor-imagery (MI) EEG decoding can exhibit subject-level failures even when average performance appears acceptable: under subject shift, a decoder can become an overconfident near-one-class predictor. This is especially problematic in source-free deployment, where target-user labels are unavailable during adaptation and expert selection. We present \textit{EEG-Fusion}, a failure-informed decision-level fusion framework that treats source-free MI decoding as label-free reliability estimation over heterogeneous experts. EEG-Fusion applies subject-wise Euclidean alignment and normalization-only test-time adaptation, then routes each target subject to a neural, covariance-based, or physiological-feature expert using a reliability gate trained on source-held-out folds to predict expert performance and collapse risk from label-free stream diagnostics. The gate uses confidence, entropy, prediction diversity, expert agreement, and predicted class balance; collapse is measured as the maximum predicted class fraction. In 9-fold leave-one-subject-out (LOSO) evaluation with three seeds, relative to a no-alignment raw EEGNet source-free anchor, EEG-Fusion improves subject macro-F1 from 0.417 to 0.529 on BCI IV-2a local protocol, from 0.314 to 0.482 on BNCI2014-001, and from 0.607 to 0.708 on BNCI2014-004; corresponding collapse-index reductions are 0.199, 0.227, and 0.169. In a 9-subject Cho2017 external subset, EEG-Fusion improves macro-F1 from 0.516 to 0.630. These results suggest that label-free reliability estimation can reduce subject-level failure modes in source-free MI-EEG deployment.