不确定数据分类的模糊随机森林算法

Handling uncertain data classification problems using fuzzy random forest

实际应用中不确定数据的分类问题越来越受到人们的重视,不确定数据不但属性值是不确定的,类标签也可能不确定。提出的不确定离散化算法,使模糊决策树能够处理区间数据,统一了属性值不确定与类标签不确定的差异。由此提出的由模糊决策树构造模糊随机森林的模糊分类算法,既继承了模糊决策树对不确定数据分类的灵活性,又继承了随机森林的集成性、鲁棒性和随机性的优点。实验结果表明,对于不确定性数据分类问题,该算法性能优于现有的一些算法。

More and more attention is paid to deal with uncertain data classification problems in many real-world applications.Values of uncertain data’s attributes and the class label are not precise.An uncertain discretization algorithm was proposed to enable the proposed fuzzy decision tree algorithm to deal with the interval data.The difference between attribute value uncertainty and class label uncertainty was unified.A classification algorithm called fuzzy random forest was proposed.The flexibility of fuzzy decision tree for uncertain data classification,the robustness of ensemble,and the power of the randomness of random forest were combined.Experimental results verify that the proposed algorithm performs better than some of the existing algorithms.