复杂决策表的特征提取方法研究

Research on the Feature Subset Selection of Complex Decision Tables

特征提取是机器学习和决策表分析的重要步骤 ,直接关系到学习和决策质量 .文中对目前特征提取的几种主要方法 ,即动态约简、神经网络、超平面、模板和作者提出的决策表分解、ITIL算法进行了评述 ,指出这些方法存在的不足 ,为特征提取的进一步研究指明了方向 .

Feature subset selection is an important topic in the fields of machine learning and rough analysis. It has the direct effect on the quality of knowledge elicited and corresponding decision. Cmade In this paper , some main methods for feature selection, including dynamic reduct ,neural networks, hyperplane plane, generalized template and the decomposition of decision tables , ITIL proposed herein are reviowed and the related problems are pointed out for further research.