Formulating Divergence Framework for Multiclass Motor Imagery EEG Brain Computer Interface

Abstract

Similar to most of the real world data, the ubiquitous presence of non-stationarities in the EEG signals significantly perturb the feature distribution thus deteriorating the performance of Brain Computer Interface. In this letter, a novel method is proposed based on Joint Approximate Diagonalization (JAD) to optimize stationarity for multiclass motor imagery Brain Computer Interface (BCI) in an information theoretic framework. Specifically, in the proposed method, we estimate the subspace which optimizes the discriminability between the classes and simultaneously preserve stationarity within the motor imagery classes. We determine the subspace for the proposed approach through optimization using gradient descent on an orthogonal manifold. The performance of the proposed stationarity enforcing algorithm is compared to that of baseline One-Versus-Rest (OVR)-CSP and JAD on publicly available BCI competition IV dataset IIa. Results show that an improvement in average classification accuracies across the subjects over the baseline algorithms and thus essence of alleviating within session nonstationarities.

Deep CNN model based on serial-parallel structure optimization for four-class motor imagery EEG classification

“…Another common issue is the stationarity preservation of MI EEG data. We may take the gradient descent on an orthogonal manifold as a meaningful reference to enforce the stationarity of EEG [38][39], which could improve the accuracy. Also, there are some points we can improve even though we made the frequency band dependent model performance evaluation.…”

An EEG-Based Brain-Computer Interface Using Spectral Correlation Function

“…Although various BCI algorithms have been examined, research on efficient feature extraction by considering the essence of EEG signals is still ongoing. According to Kumar et al [72], the almost constant presence of nonstationarities in EEG signals degrades BCI performance. In an information theoretic context, the authors suggested a novel strategy based on Joint Approximate Diagonalization (JAD) to optimize stationarity for multiclass motor imagery BCIs.…”

E‐CNNet: Time‐reassigned Multisynchrosqueezing transform‐based deep learning framework for MI‐BCI task classification

“…Thus, BCI has significant applications in various fields such as automobiles, robotics and is widely used in medical treatment. 3 To monitor the brain activities, numerous techniques such as Electrocorticography (ECoG), functional magnetic resonance imaging (fMRI), magneto-encephalography (MEG), and electroencephalogram (EEG) are used. 4 Since EEG is non-invasive, portable, and low in cost, it is one of the widely used neuroimaging technology to record the brain activities.…”