Exploiting diversity for optimizing margin distribution in ensemble learning
作者：
Highlights：
• Double Rotation is proposed to produce diversity among base classifiers.
• We construct a margin related loss function to learn the weights of base classifiers.
• Margin Based Pruning is proposed to improve margin distribution of ensembles.
• Extensive experiments are conducted to validate the effectiveness of DRMP.
• We explain the rationality of DRMP from different perspectives.
摘要
•Double Rotation is proposed to produce diversity among base classifiers.•We construct a margin related loss function to learn the weights of base classifiers.•Margin Based Pruning is proposed to improve margin distribution of ensembles.•Extensive experiments are conducted to validate the effectiveness of DRMP.•We explain the rationality of DRMP from different perspectives.
论文关键词：Ensemble learning,Margin distribution,Diversity,Fusion strategy,Rotation
论文评审过程：Received 9 December 2013, Revised 29 March 2014, Accepted 9 June 2014, Available online 18 June 2014.
论文官网地址：https://doi.org/10.1016/j.knosys.2014.06.005
