.png){kind=icon}
返回
Cluster decomposing and multi-objective optimization based-ensemble learning framework for motor imagery-based brain-computer interfaces
.png){kind=icon}
PRE
AI
DOI：10.1088/1741-2552/abe20f.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Objective. Motor imagery (MI) is a mental representation of motor behavior and a widely used pattern in electroencephalogram (EEG) based brain-computer interface (BCI) systems. EEG is known for its non-stationary, non-linear features and sensitivity to artifacts from various sources. This study aimed to design a powerful classifier with a strong generalization capability for MI based BCIs. Approach. In this study, we proposed a cluster decomposing based ensemble learning framework (CDECL) for EEG classification of MI based BCIs. The EEG data was decomposed into sub-data sets with different distributions by clustering decomposition. Then a set of heterogeneous classifiers was trained on each sub-data set for generating a diversified classifier search space. To obtain the optimal classifier combination, the ensemble learning was formulated as a multi-objective optimization problem and a stochastic fractal based binary multi-objective fruit fly optimization algorithm was proposed for solving the ensemble learning problem. Main results. The proposed method was validated on two public EEG datasets (BCI Competition IV datasets IIb and BCI Competition IV dataset IIa) and compared with several other competing classification methods. Experimental results showed that the proposed CDECL based methods can effectively construct a diversity ensemble classifier and exhibits superior classification performance in comparison with several competing methods. Significance. The proposed method is promising for improving the performance of MI-based BCIs.
Keyword:
brain-computer interface
motor imagery
ensemble learning
multi-objective fruit fly optimization algorithm
期刊
IF:
3.8
论文数:
4.0K
被引数:
1.4W
机构
暂无机构信息
引用论文
Internal Feature Selection Method of CSP Based on L1-Norm and Dempster-Shafer Theory基于L1-Norm和dempster-shafer理论的CSP内部特征选择方法
OAAI
OAAI分享收藏Channel selection by Rayleigh coefficient maximization based genetic algorithm for classifying single-trial motor imagery EEG.png)
NEUROCOMPUTING
NEUROCOMPUTINGIF6.5
PREAI
分享收藏Correlation-based channel selection and regularized feature optimization for MI-based BCINEURAL NETWORKS
NEURAL NETWORKSIF6.3
PREAI
分享收藏
查看更多内容