S. R. Liyanage, Cuntai Guan, Haihong Zhang, Kai Keng Ang, Jian-Xin Xu and Tong Heng Lee, "Dynamically Weighted Classification with Clustering to tackle non-stationarity in Brain computer Interfacing," The 2012 International Joint Conference on Neural Networks (IJCNN), Brisbane, QLD, 2012, pp. 1-6. doi: 10.1109/IJCNN.2012.6252652
The performance of Brain-Computer Interface (BCI) applications are sometimes hindered by non-stationarity in the EEG data. This paper proposes a novel Dynamically Weighted Classification with Clustering (DWCC) to handle the non-stationarity in the EEG data. The proposed method employs cosine distance based and mutual Information-based criteria in a k-means clustering framework. The proposed method addresses the non-stationarity in EEG since non-stationary sources form clusters in the feature space that are distinctly apart from one another. Multiple classifiers are trained independently based on these clustered features. A dynamically-weighted classifier ensemble approach is proposed to perform the classification. The dynamic weights are based on the distances from a testing sample to cluster centres. Experimental results on publicly available BCI Competition IV Dataset 2a yielded a mean accuracy of 81.5% which is statistically significant (t-test p<0.05) compared to the baseline result of 75.9% using a single classifier.