特征选择算法研究综述

A Research Review of Feature Selection Algorithm

模式识别领域中,特征选择作为预处理模块的关键步骤,特征选择函数用来降低特征空间的维数,提高分类器的分类性能。首先分析了特征选择的主要过程,从不同视角探讨了特征选择的分类方法,然后分析了基于SVM进行特征选择的作用,对基于SVM的特征选择算法进行了归纳总结,最后分析对比了基于SVM的三类特征选择的优缺点,指出理论研究和实际应用中的研究热点和应用发展方向。

Feature selection is the key step of the preprocessing module in pattern recognition,which is used to reduce the dimension of feature space and improve the classification performance of the classifier.First,the paper analyzes the main process of feature selection,and discusses the classification method of feature selection from different perspectives.Then the paper analyzes the function of feature selection based on SVM,and summarizes the feature selection algorithm based on SVM.Finally,the paper analyzes and compares the advantages and disadvantages of three kinds of feature selection based on SVM,and aslo points out the research hotspots and application development directions in theoretical research and practical application.