论文部分内容阅读
基于邻域粗糙集以及模糊粗糙集等价关系下的属性约简方法,引入α信息熵,建立模糊相似关系下的α信息熵不确定性度量,提出基于α信息熵的属性重要度度量,并以此构建混合属性约简算法.利用UCI数据集与几种相关的约简方法进行比较,验证了该方法可以选择较少属性的同时保证较高的分类精确性.实际应用中,对参数α的有效调节,可获得多个约简结果,进而可根据需要选择最佳约简.
Based on the attribute reduction method of neighborhood rough set and fuzzy rough set equivalence, the α information entropy is introduced to establish the uncertainty measure of α information entropy under the fuzzy similarity relation, and the attribute importance measure based on α information entropy is proposed. In this way, a hybrid attribute reduction algorithm is constructed.Comparing the UCI dataset with several related reduction methods, it is verified that this method can select fewer attributes and ensure higher classification accuracy.In practice, The effective adjustment, you can get a number of reduction results, which can be selected according to the best reduction.