稳定的特征选择研究

Research on stable feature selection

特征选择是机器学习和数据挖掘领域的关键问题之一,而特征选择的稳定性也是目前的一个研究热点。主要对特征选择的稳定性因素和稳定性度量进行分析,并详细介绍了目前比较经典的两种提高特征选择稳定性的方法。

Feature selection is one of the key problems in machine learning and data mining, and the stability of feature selection is one of the current hot points. In the paper, the factors to the stability of feature selection and the measure of stability is introduced, at the same time, two classic methods to improve the stability is presented.