WCSE

Classification and Recognition of EEG Signals Based on AdaBoost Algorithm

Abstract— EEG signals directly and objectively reflect the activity of the human body, and are widely used in the field of human movement intention recognition. In order to realize the recognition of the movement intention, firstly, the ERD/ERS phenomenon that occurs when the human body performs a certain side movement or imaginary movement is determined by three methods based on autoregressive model coefficients, based on wavelet packet decomposition coefficient energy and based on co-space mode. Perform feature extraction. After feature extraction, in order to achieve classification, an ensemble learning classification method combining linear discriminant analysis (LDA) and adaptive lifting method is proposed. LAD is used as the weak classifier of AdaBoost to construct a strong classifier for recognition. Compared with the single classification method of the ordinary LDA algorithm, it is proved that the method of using AdaBoosting to strengthen the learning makes the classification accuracy of the motor imagery EEG signal significantly improved. The experimental results show that the use of AdaBoosting-LDA for EEG signal fusion features can achieve a maximum accuracy of 90.7%.

Index Terms— EEG signal, Feature extraction, LDA, AdaBoosting

Yanran Wang
School of AutomationWuhan University of Technology
Zhijun Li
School of AutomationWuhan University of Technology
Rui Zhang
School of AutomationWuhan University of Technology
Keyong Deng
CCCC Second Harbor Consultants Co,Ltd

[Download]

Cite: Yanran Wang, Zhijun Li, Rui Zhang, Keyong Deng, " Classification and Recognition of EEG Signals Based on AdaBoost Algorithm, " WCSE 2022 Spring Event: 2022 9th International Conference on Industrial Engineering and Applications, pp. 1447-1455, Sanya, China, April 15-18, 2022.