Transferring Common Spatial Filters With Semi-Supervised Learning for Zero-Training Motor Imagery Brain-Computer Interface

Common spatial pattern (CSP) has been proved to be one of the most efficient feature-extracting methods for brain–computer interfaces (BCIs), especially for motor imagery BCI. However, CSP is a supervised method and performs poorly when there are not enough labeled data. This paper aims to construct a minimum-training BCI, which means there are only a few labeled data, even none labeled data for target subjects. Under this condition, conventional CSP cannot work well. Therefore, source data (related labeled data from other subjects) are exploited and common filters across subjects are obtained using a clustering method. After that, features are extracted and a semi-supervised support vector machine which transfers knowledge across subjects is proposed. The experiments illustrate the effectiveness of our algorithm. When there are none labeled data for target subjects, our algorithm outperforms two state-of-the-art algorithms in semi-supervised learning field, and as the amount of unlabeled data for target subjects becomes larger, the performance of our algorithm grows better and better, which is suitable for online use. When the amount of labeled data for target subjects (denoted as <inline-formula> <tex-math notation="LaTeX">$M$ </tex-math></inline-formula> in this paper) is small, our algorithm also shows its strength compared with corresponding outstanding algorithms. For dataset IVa of BCI competition III, our algorithm performs the best for all the subjects excluding “aw” when <inline-formula> <tex-math notation="LaTeX">${M}=20$ </tex-math></inline-formula>. Compared with counterparts, our method outperforms 6.6% for “aa,” 19.3% for “al,” 11.8% for “av,” and 9.7% for “ay” on average, respectively. When <inline-formula> <tex-math notation="LaTeX">${M}=40$ </tex-math></inline-formula>, our method performs the best for “al” and “av”. It averagely outperforms 8.0% for “al” and 15.3% for “av,” respectively. For GigaDataset, averagely our method outperforms 4.4% for “sbj2,” 5.0% for “sbj4,” and 7.1% for “sbj5” when <inline-formula> <tex-math notation="LaTeX">${M}=20$ </tex-math></inline-formula>, respectively. When <inline-formula> <tex-math notation="LaTeX">${M}=40$ </tex-math></inline-formula>, our algorithm performs the best only for “sbj5.” It averagely outperforms 8.6% compared with counterparts. Although the performances of our method are not the best for all the conditions, its performances are very robust and competitive.