基于FastICA和SVM的EEG信号分类系统

An EEG Signal System Based on FastICA and SVM

针对大量EEG数据进行分析时,视觉检测显得既费时效率又低,提出了EEG信号分类系统.该系统采用FastICA方法获取EEG信号模式的高阶统计信息,并将输入模式空间映射到相应的独立成分空间,然后利用SVM在独立成分空间中构造广义最优分类超平面.实验研究结果表明:系统综合了FastICA和SVM特性,具有响应实时、漏检率低等优点.

An EEG signal detection ensemble system for solving the low rate of vision detection is developed when analysis so many EEG signals. A novel FastICA method was presented, in which the independent component analysis approach was used to acquire the high order statistic infor- mation of EEG intrusion action mode and mapped the input mode space in to the corresponding in- dependent component space. Then the generalized maximal margin hyperplane was constructed in the independent component space using the support vector machine. Testing results show that the system integrates the features of FastICA and SVM to response real-time and lower the rate of false negative.