Decoding EEG in Motor Imagery Tasks with Graph Semi-Supervised Broad Learning

She She,Zhou Zhou,Gan Gan,Ma Ma,Luo Luo

doi:10.3390/electronics8111273

Abstract

In recent years, the accurate and real-time classification of electroencephalogram (EEG) signals has drawn increasing attention in the application of brain-computer interface technology (BCI). Supervised methods used to classify EEG signals have gotten satisfactory results. However, unlabeled samples are more frequent than labeled samples, so how to simultaneously utilize limited labeled samples and many unlabeled samples becomes a research hotspot. In this paper, we propose a new graph-based semi-supervised broad learning system (GSS-BLS), which combines the graph label propagation method to obtain pseudo-labels and then trains the GSS-BLS classifier together with other labeled samples. Three BCI competition datasets are used to assess the GSS-BLS approach and five comparison algorithms: BLS, ELM, HELM, LapSVM and SMIR. The experimental results show that GSS-BLS achieves satisfying Cohen’s kappa values in three datasets. GSS-BLS achieves the better results of each subject in the 2-class and 4-class datasets and has significant improvements compared with original BLS except subject C6. Therefore, the proposed GSS-BLS is an effective semi-supervised algorithm for classifying EEG signals.

Highlights

The Brain-Computer Interface (BCI) is a technology that only needs to use the signals generated by the human brain when subjected to specific stimuli to control external devices or systems [1], which is independent of normal peripheral neuromuscular channels
The graph-based semi-supervised broad learning system (GSS-broad learning system (BLS)) was superior to other algorithms when the ratio was less than 50% while Laplacian SVM (LapSVM) was better when it was above 50% in Generally, the results showed that the GSS-BLS outperformed the other algorithms in small training samples since the GSS-BLS exploits the underlying manifold structure of the labeled and unlabeled data space
The graph constructed for LapSVM was better in three subjects, and the original EEG data is preprocessed so we might reserve the main information of motor imagery including artifacts

Summary

Introduction

The Brain-Computer Interface (BCI) is a technology that only needs to use the signals generated by the human brain when subjected to specific stimuli to control external devices or systems [1], which is independent of normal peripheral neuromuscular channels. The application of BCI technology has become more and more extensive, which has achieved fruitful results in the fields of games, rehabilitation, and aerospace [2]. BCI is mainly used to accurately detect the patient’s intention of exercise in the field of active motor rehabilitation, so the patients can actively participate in the process of exercise training and induce neural plasticity [3]. This is due to the low cost of electroencephalogram (EEG) signals acquisition, ease of use, and minimal side effects in the subjects. It remains a challenging task to achieve accurate and real-time classification of EEG signals

Methods

Results

Discussion

Conclusion