Adversarial Feature Learning in Brain Interfacing: An Experimental Study on Eliminating Drowsiness Effects
This work addresses a critical issue for users of EEG-based BCIs, such as those with disabilities, by reducing performance variability, though it is incremental as it builds on existing deep learning architectures.
The study tackled the problem of performance fluctuations in EEG-based brain-computer interfaces due to drowsiness and fatigue during long recordings, and demonstrated that adversarial invariant feature learning can effectively eliminate these effects, achieving a 15% improvement in classification accuracy compared to baseline methods.
Across- and within-recording variabilities in electroencephalographic (EEG) activity is a major limitation in EEG-based brain-computer interfaces (BCIs). Specifically, gradual changes in fatigue and vigilance levels during long EEG recording durations and BCI system usage bring along significant fluctuations in BCI performances even when these systems are calibrated daily. We address this in an experimental offline study from EEG-based BCI speller usage data acquired for one hour duration. As the main part of our methodological approach, we propose the concept of adversarial invariant feature learning for BCIs as a regularization approach on recently expanding EEG deep learning architectures, to learn nuisance-invariant discriminative features. We empirically demonstrate the feasibility of adversarial feature learning on eliminating drowsiness effects from event related EEG activity features, by using temporal recording block ordering as the source of drowsiness variability.