Genetic algorithms based LSSVM for EEG fatigue multi-classification
编号:121
访问权限:仅限参会人
更新:2022-05-19 16:19:01
浏览:195次
张贴报告
摘要
In order to improve the classification accuracy of multiclassification of EEG fatigue data, a genetic algorithm based least squares support vector machine algorithm (GA-LSSVM) is proposed in this paper. Firstly, the hyperparameters σ (kernel function width) and γ (the regularization parameter) of LSSVM are optimized by GA to obtain the GA-LSSVM algorithm model. Secondly the electroence-phalographic (EEG) signals use SEED-VIG fatigue data with 17 channels five frequency bands and 4 features. The data are divided into training set and test set according to 7:3 proportion to train SVM model and verify the presented algorithm. Experiments evidence that the GA-LSSVM algorithm improves the classification accuracy of EEG signals.
发表评论