EEG Signal Classification Using Manifold Learning and Matrix‐Variate Gaussian Model

Lei Zhu,Junting Yang,Jianhai Zhang,Qifeng Hu,Ping Xu,Nanjiao Ying

doi:10.1155/2021/6668859

Abstract

In brain-computer interface (BCI), feature extraction is the key to the accuracy of recognition. There is important local structural information in the EEG signals, which is effective for classification; and this locality of EEG features not only exists in the spatial channel position but also exists in the frequency domain. In order to retain sufficient spatial structure and frequency information, we use one-versus-rest filter bank common spatial patterns (OVR-FBCSP) to preprocess the data and extract preliminary features. On this basis, we conduct research and discussion on feature extraction methods. One-dimensional feature extraction methods like linear discriminant analysis (LDA) may destroy this kind of structural information. Traditional manifold learning methods or two-dimensional feature extraction methods cannot extract both types of information at the same time. We introduced the bilinear structure and matrix-variate Gaussian model into two-dimensional discriminant locality preserving projection (2DDLPP) algorithm and decompose EEG signals into spatial and spectral parts. Afterwards, the most discriminative features were selected through a weight calculation method. We tested the method on BCI competition data sets 2a, data sets IIIa, and data sets collected by our laboratory, and the results were expressed in terms of recognition accuracy. The cross-validation results were 75.69%, 70.46%, and 54.49%, respectively. The average recognition accuracy of new method is improved by 7.14%, 7.38%, 4.86%, and 3.8% compared to those of LDA, two-dimensional linear discriminant analysis (2DLDA), discriminant locality property projections (DLPP), and 2DDLPP, respectively. Therefore, we consider that the proposed method is effective for EEG classification.

Highlights

Brain-computer interface (BCI) is a kind of real-time communication system connecting the brain and external devices
BCIs based on electroencephalogram (EEG) can convert the information sent by the brain into commands to drive the external devices, so as to realize the communication between people and the outside world [1]. ere are several control signal types in BCI and, among them, motor imagery (MI) is one of the most studied applications [2]
To solve the above issues and combine the important local information in the EEG signal with the 2D matrix processing method, we propose an extension of 2DDLPP based on the Gaussian variable model. e main idea of matrix-variable Gaussian model [28] implies a separable structure for the covariance matrix of the vectorized data and it shows that the covariance between any two spatialspectral features can be decomposed into two terms

Summary

Introduction

Brain-computer interface (BCI) is a kind of real-time communication system connecting the brain and external devices. Studies on EEG signal indicate that when people perform motor imaging tasks, this will cause an event-related desynchronization (ERD) and event-related synchronization (ERS) of oscillations in alpha band (8–13 Hz) and beta band (14–30 Hz) [3]. Due to these characteristics, researchers can process and analyze EEG signals in relevant frequency bands for the classification of motor imaging tasks. To solve the computational complexity and data storage problem caused by the high dimension of signals, many dimensionality reduction methods have been used in traditional BCI technology. PCA seeks to learn a projection that can preserve the main energy of data, it does not contain

Methods

Results

Conclusion