【摘要】 提出一种新的基于核Fisher判别分析的意识任务识别新方法。该方法首先通过核函数建立一个非线性映射,把原空间的样本点投影到一个高维特征空间,然后在特征空间应用线性Fisher判别。利用不同意识任务生成的脑电数据对KFDA和FDA进行比较,最后用线性支持向量机进行分类和识别,并与非线性支持向量机进行了比较,结果表明KFDA的识别率明显优于后二者。更多还原

【Abstract】 A new mental task recognition method based on kernel Fisher discriminant is proposed.The method is based on a non-linear mapping from input space to feature space using kernel function.Linear Fisher discriminant is used in feature space.KFDA-based mental task recognition method is tested and compared with FDA and non-linear SVM using the same EEG signals.Experiment results indicate that the performance of KFDA-based mental task recognition method is superior to the others.更多还原