集成学习之随机森林算法综述  

A Survey of Random Forests Algorithms

作　　者：王奕森 夏树涛[1] 

机构地区：[1]清华大学深圳研究生院/计算机科学与技术系,深圳518055

出　　处：《信息通信技术》2018年第1期49-55,共7页Information and communications Technologies

基　　金：国家自然科学基金资助项目(61771273)

摘　　要：集成学习是一类非常重要且实用的方法,以简单高效著称的随机森林算法是集成学习算法的代表之一,它集成众多决策树并以投票的方式输出结果,在许多应用领域取得了巨大的成功。文章介绍决策树和随机森林算法,总结随机森林算法在性能改进、理论性质方面的研究进展,及其和深度学习算法之间的区别与联系。Ensemble learning is a very important and practical method. Random forests algorithm is one most famous representative among ensemble learning methods, which combines a number of decision trees and votes for the fnal prediction, leading to a great success in many real-world tasks. This paper briefy reviews the decision tree and random forests algorithms, and summarizes the progress in empirical performance and theoretical properties of random forests algorithm, and the difference and relation between random forests and the deep learning methods.

关 键 词：随机森林 机器学习 深度学习 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]