随机决策树方法总结
首发时间:2010-11-29
摘要:随机决策树(Random Decision Tree)作为一种不含有任何属性选择的决策树构建方法,不仅可以作为一种分类和回归方法使用,还可以作为集成学习中的基学习器使用。集成学习的基学习器的选择除了要保证每个基学习器尽量的不同,又要保证每个基学习器的学习能力尽量好。不同学者对随机决策树方法进行改进,利用其随机化的特点达到多样性,又通过其他方式来控制这种随机性,使算法在随机性和判定性之间保持相对平衡,以达到最佳的学习效果。最后介绍了随机决策树的一些应用。
For information in English, please click here
Review on Random Decision Tree
Abstract:Random Decision Tree, a tree construction method without an optimization process, not only can serve as classification/regression tool but also can be used as base learner in ensemble learning. In a good ensemble learning configuration, the base learners should be as diverse as possible, and each individual should have a strong learning ability, which makes it really hard to be strong at both criteria. Several methods had been proposed to improve RDT. The basic idea is to utilize the randomness to create diversity, at the same time some methods are proposed to control the degree of the randomness, adding some deterministic to the randomness, in order to get a better learner. In the end, several applications of RDT were reviewed.
Keywords: Random decision tree Classification Ensemble learning
论文图表:
引用
No.4390436550045129****
同行评议
共计0人参与
勘误表
随机决策树方法总结
评论
全部评论0/1000