In view of the problems of low accuracy with filter feature selection algorithm and long train-ing time of wrapper feature selection algorithm,a hybrid feature selection algorithm based on mixed information gain and adaptive genetic algorithm was proposed.The filter bank common spatial pattern was used to extract the features of motor imagery electroencephalogram signals,and the information gain of each feature calculated and sorted.According to the sorting,the threshold value method was used to eliminate some useless features.The optimal feature subset sought out from the remaining features was classified via the adaptive genetic algorithm.Two common data sets were used to verify the validity of the algorithm,and the average classification accuracy was 81.24％±15.04％and the average time was 3.68 s.The experimental results showed that the classification accuracy of the proposed algorithm was higher than that of the filter algorithm,and the training time was shorter than that of the wrapper algorithm.
motor imageryfeature selectioninformation gainadaptive genetic algorithm
维普
万方数据
