TY - GEN
T1 - Design of a Cascade Stacking Ensemble Algorithm Based on Multi-domain Feature Fusion of MI-EEG
AU - Zhang, Yue
AU - Chen, Weihai
AU - Lin, Chu Liang
AU - Pei, Zhongcai
AU - Chen, Jianer
AU - Shi, Hongfei
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Electroencephalogram (EEG) signal is a complex bioelectrical signal with non-stationary, nonlinear, high dimension, and low signal-to-noise ratio. Single time domain or frequency domain analysis has great limitations, and the wavelet transform combined with time-frequency analysis cannot completely distinguish the artefact components like the EEG signals. All these problems will reduce the classification recognition rate. In view of this, an ensemble learning algorithm based on multi-domain features is proposed to realize the classification of MI-EEG. Four complementary features of time domain, frequency domain, spatial domain and time-frequency feature are calculated and fused as feature vectors, and then classified by an ensemble algorithm. Then, a cascade classification model based on a Stacking integrated algorithm is proposed. The first level is composed of Random Forest (RF), Gradient Boosting Decision Tree (GBDT) and Extreme Gradient Boosting (XGBoost) model, and the second level is composed of logistic regression (LR) to output the results. The output results show that the classification results of fusion features are significantly improved compared with those of single features. the accuracy of the Stacking integrated algorithm is up to 90.12%. Compared with a single classification algorithm, the average classification accuracy is improved by 3%.
AB - Electroencephalogram (EEG) signal is a complex bioelectrical signal with non-stationary, nonlinear, high dimension, and low signal-to-noise ratio. Single time domain or frequency domain analysis has great limitations, and the wavelet transform combined with time-frequency analysis cannot completely distinguish the artefact components like the EEG signals. All these problems will reduce the classification recognition rate. In view of this, an ensemble learning algorithm based on multi-domain features is proposed to realize the classification of MI-EEG. Four complementary features of time domain, frequency domain, spatial domain and time-frequency feature are calculated and fused as feature vectors, and then classified by an ensemble algorithm. Then, a cascade classification model based on a Stacking integrated algorithm is proposed. The first level is composed of Random Forest (RF), Gradient Boosting Decision Tree (GBDT) and Extreme Gradient Boosting (XGBoost) model, and the second level is composed of logistic regression (LR) to output the results. The output results show that the classification results of fusion features are significantly improved compared with those of single features. the accuracy of the Stacking integrated algorithm is up to 90.12%. Compared with a single classification algorithm, the average classification accuracy is improved by 3%.
KW - Brain-computer interface
KW - EEG
KW - Ensemble learning
UR - https://www.scopus.com/pages/publications/85146901856
U2 - 10.1109/ICIEA54703.2022.10006088
DO - 10.1109/ICIEA54703.2022.10006088
M3 - 会议稿件
AN - SCOPUS:85146901856
T3 - ICIEA 2022 - Proceedings of the 17th IEEE Conference on Industrial Electronics and Applications
SP - 931
EP - 936
BT - ICIEA 2022 - Proceedings of the 17th IEEE Conference on Industrial Electronics and Applications
A2 - Xie, Wenxiang
A2 - Gao, Shibin
A2 - He, Xiaoqiong
A2 - Zhu, Xing
A2 - Huang, Jingjing
A2 - Chen, Weirong
A2 - Ma, Lei
A2 - Shu, Haiyan
A2 - Cao, Wenping
A2 - Jiang, Lijun
A2 - Shu, Zeliang
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 17th IEEE Conference on Industrial Electronics and Applications, ICIEA 2022
Y2 - 16 December 2022 through 19 December 2022
ER -