The comparison of automatic artifact removal methods with robust classification strategies in terms of EEG classification accuracy

One of the key objectives of brain-computer interface (BCI) design is to construct accurate electroencephalogram (EEG) based classifier. But out of laboratory all EEG signals are contaminated with artifacts, which hamper algorithmic processing and EEG analysis, i.e. classifier ought to get a prediction for noisy data. Real-time BCI system rely on relatively clean EEG signals. Therefore, the exclusion of artifacts is of special interest for BCI applications in everyday life. There are two main approaches to this objective: automatic EEG artifact rejection methods (subtract the noisy component) and robust classification methods (replace sensitive to outliers estimates with robust counterparts). The goal of this work is to quantitatively compare popular automatic EEG artifact rejection approaches with robust classification methods in terms of motor imagery (MI) classification paradigm.