Fuzzy transfer learning approach for analysing imagery BCI tasks

In brain-computer interfaces (BCI), the statistical distribution of the data could differ across subjects as well as across sessions for an individual subject. Moreover, the lack of data due to the difficulties in collecting data in BCI is a common challenge in training the systems. Since most of machine learning tools are based on the assumption that the distribution of training and testing data are the same and they need adequate training data, they would fail in such situations. To overcome this problem and because of the vague and uncertain essence of EEG data, in this paper, we used a fuzzy transfer learning (FTL) method based on Generalized Hidden-Mapping Ridge Regression (GHRR) to improve the classification task in BCI. Takagi-Sugeno-Kang fuzzy logical system (TSK) with proposed modified Wang-Mendel fuzzy rule generation were employed for classification. Then the session-to-session transfer of knowledge is adopted. The results demonstrate the effectiveness of our proposed method in classification and outperform the well-known SVM classifier.