摘要
演化决策树方法将传统的决策树算法与演化算法相结合,具有全局搜索的优点。基于集成学习框架,提出了Adaboost演化决策树算法,并对基本遗传算子加以改进。实验结果表明Adaboost演化决策树能在较短的演化代数内得到较高的预测准确度。
Evolutionary decision tree method has the advantage of global search. Based on the framework of ensembel learning, this paper proposes an algorithm of Adaboost evolutionary decision tree in which genetic operators are also improved. The experimental results show that this method can reach higher prediction accuracy in a smaller number of evolution generations.
出处
《计算机应用与软件》
CSCD
北大核心
2007年第3期1-2,21,共3页
Computer Applications and Software
基金
国家自然科学基金(60005004)
安徽省自然科学基金(01042302)资助
关键词
决策树
演化算法
集成学习
Decision tree Evolutionary computation Ensemble learning
