Some existing feature subset selection algorithms use only one metric, such as symmetric uncertainty, to evaluate redundant features. These algorithms have the problem that some relevant features are considered as redundant and removed. To solve the problem, this paper proposes a feature subset selection algorithm based on equal interval division and three-way interaction information. Symmetric uncertainty between each feature and the class label is first calculated and compared with zero, and irrelevant features are removed. Then, symmetric uncertainty between features is calculated and compared with symmetric uncertainty between features and the class label. The method of equal interval division and ranking is adopted to process symmetric uncertainty between features and the class label as well as symmetric uncertainty between features, and these processed parts are then compared. Three-way interaction information among features and the class label is calculated and compared with zero. The three results of the comparisons are employed to remove redundant features. To validate the performance, the proposed algorithm is compared with several feature subset selection algorithms. Experimental results demonstrate that the proposed algorithm can achieve better feature selection performance.

一些现有的特征子集选择算法仅使用一种度量（例如对称不确定性）来评估冗余特征。这些算法的问题是一些相关的特征被认为是冗余的并被删除。针对该问题，本文提出了一种基于等距划分和三路交互信息的特征子集选择算法。首先计算每个特征与类标签之间的对称不确定性并与零进行比较，然后去除不相关的特征。然后，计算特征之间的对称不确定性，并与特征和类标签之间的对称不确定性进行比较。采用等距划分和排序的方法来处理特征与类标签之间的对称不确定性以及特征之间的对称不确定性，然后比较这些加工过的零件。计算特征与类标签之间的三向交互信息并与零进行比较。比较的三个结果用于去除冗余特征。为了验证性能，将所提出的算法与几种特征子集选择算法进行了比较。实验结果表明，该算法能够取得较好的特征选择性能。