基于K近邻分类间隔的特征选择方法研究

Feature Selection Based on Margin of K-Nearest Neighbors

特征选择是机器学习和模式识别领域的一个关键问题。文中详细分析研究一类基于K近邻分类间隔的特征选择算法,并着重讨论当K>1时,特征选择的评价准则和搜索策略的设计,同时在多个数据集上验证其性能。

Feature selection is one of key problems in machine learning and pattern recognition.In this paper,a type of feature selection methods based on Margin of K-nearest neighbors is discussed.Furthermore,the feature selection evaluation criterion and search strategy is introduced when the value of k is more than 1.Meanwhile,the experimental results on different data sets are presented.