集成学习中特征选择技术

集成学习和特征选择是当前机器学习领域中的研究热点.集成学习通过重复采样可产生个体学习器之间差异度，从而提高个体学习器的泛化能力，特征选择应用到集成学习可进一步提高集成学习技术的效果，该研究有3个方面：数据子集的特征选择、个体学习器的选择和多任务学习.该文对近几年集成学习中特征选择技术的研究进行回顾，尤其对以上3个方面的研究分别进行总结，提出一些共性的技术指导以后的研究.

Feature Selection for Ensemble Learning

Ensemble learning and feature selection are hot topics in machine learning studies. The improvement of generalization performance of individuals comes primarily from the diversity caused by re-sampling the training set. Feature selection for ensemble learning can also improve diversity in three aspects: feature selection for individuals, selective ensemble learning, and multi task learning. This paper gives an overview of feature selection methods for ensemble learning in recent years, and summarize some general techniques useful in the further studies.