A hierarchical semi-supervised extreme learning machine method for EEG recognition

Abstract

Feature extraction and classification is a vital part in motor imagery-based brain-computer interface (BCI) system. Traditional deep learning (DL) methods usually perform better with more labeled training samples. Unfortunately, the labeled samples are usually scarce for electroencephalography (EEG) data, while unlabeled samples are available in large quantity and easy to collect. In addition, traditional DL algorithms are notoriously time-consuming for the training process. To address these issues, a novel method of hierarchical semi-supervised extreme learning machine (HSS-ELM) is proposed in this paper and applied for motor imagery (MI) task classification. Firstly, the deep architecture of hierarchical ELM (H-ELM) approach is employed for feature learning automatically, and then these new high-level features are classified using the semi-supervised ELM (SS-ELM) algorithm which can exploit the information from both labeled and unlabeled data. Extensive experiments were conducted on some benchmark datasets and EEG datasets to evaluate the effectiveness of the proposed method. Compared with several state-of-the-art methods, including SVM, ELM, SAE, H-ELM, and SS-ELM, our HSS-ELM method can achieve better classification accuracy, a mean kappa value of 0.7945 and 0.5701 across all subjects in the training and evaluation sessions of BCI Competition IV Dataset 2a, respectively. Finally, it comes to the conclusion that the proposed method has achieved superior performance for feature extraction and classification of EEG signals. Graphical abstract The schematic of the proposed HSS-ELM algorithm.

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

“…al. [22,23] improve the ELM algorithm and propose semi-supervised ELM and safe semi-supervised ELM for EEG signals classification respectively. The results show that classification accuracy has significant improvement over ELM.…”

“…Jia et al [21] propose a new semi-supervised deep learning algorithm combined with the restricted Boltzmann machine and apply it for EEG signals classification. She et al [22,23] improve the ELM algorithm and propose semi-supervised ELM and safe semi-supervised ELM for EEG signals classification respectively. The results show that classification accuracy has significant improvement over ELM.…”

Motor Imagery Classification for Brain Computer Interface Using Deep Metric Learning

“…Most previous BCI studies used relatively small to moderate training sample size (~8-30 subjects) and used all subjects' training data to train one model (training set usually divided into 80% of each user's EEG data and the remaining was held out for testing). Then, the accuracy of this trained model has been reported on each user's test data separately [28], [41], [45], [64], [66], [71], [72]. The motivation in the previous studies to train a single model on all users' training data, while the model was tested on each individual separately was likely, because the classification performance of most classifiers usually increase with more training data.…”

“…Several studies [1], [7], [28], [32], [40], [42], [44], [45], [50], [51], [62], [66], [71], [72], [77] have investigated the classification of MI-EEG to develop a BCI system that can provide feedback during MI training and eventually use the BCI to enhance the life of patients with disabilities and paralysis.…”

“…Various methods for feature extraction have been tested in the literature. A common method of EEG feature extraction is to use Fast Fourier Transform to convert EEG signals into the frequency domain and using the Common Spatial Pattern (CSP) algorithm [28], [42], [44], [66], [77] to classify MI signals. The extracted features were used for training a classifier, where most researchers have applied classical machine learning classifiers such as Mahalanobis Linear Discrimination classifier [32], Support Vector Machine [28], [77], Bayesian classifier [50], Bayesian Linear Discriminant Analysis [42], Logistic Regression [40] or Linear Discriminant Analysis [51].…”