A Feature Gene Selection Method Based on ReliefF and PSO

Due to the shortcomings of low efficiency and low accuracy of DNA microarray data, the paper proposes a feature gene selection algorithm based on ReliefF and PSO (RefFPSO). Firstly, ReliefF is used as the feature pre-filter to delete the genes with low correlation with the classification target. Then PSO is used as the search algorithm. Finally, the classification accuracy of SVM is used as the evaluation function of the feature subset to get the final optimal gene subset. Experiments show that this method can effectively put forward irrelevant genes and use fewer characteristic genes to obtain higher classification accuracy.