论文部分内容阅读
现有基于显露模式的分类方法主要通过精简显露模式的数量以构建实用的轻量级分类器,然而对显露模式集的过度精简会损害数据信息的完整性,进而影响分类器性能.本文提出LLEP分类器,采用懒惰学习策略,将分类器的构建推迟到分类阶段进行,以在获知待分类事务信息的基础上,构建出更具针对性的局部分类器;对于显露模式的冗余消除问题,采用了等价类方法来快速划分包含重复信息的显露模式,以保留鲁棒性更优的显露模式参与分类.本文在UCI机器学习库27个数据集上的实验表明,LLEP分类器同11经典种分类器相比,在分类准确度上表现出了良好的性能.
However, the exaggeration of the explicit pattern set will impair the integrity of data information and thus affect the performance of the classifier.In this paper, we propose that LLEP The classifier uses lazy learning strategies to delay the construction of classifiers to the classification stage so as to construct a more targeted local classifier based on the information of the to-be classified transactions. For the redundancy elimination problem of explicit modes, Equivalence class method is used to quickly classify the revealing patterns containing repetitive information and retain the exposed pattern with better robustness.In this paper, experiments on 27 datasets of UCI machine learning library show that LLEP classifier is similar to 11 classic Compared with the classifier, the classification accuracy shows good performance.