Considering diversity and accuracy simultaneously for ensemble pruning

Open Webpage

Qun Dai, Rui Ye, Zhuan Liu

Published: 2017, Last Modified: 07 Nov 2024Appl. Soft Comput. 2017EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•SDAcc, DFTwo and AccRein are developed for pruning the ensemble.•SDAcc makes maximum use of all samples to further improve ensemble performance.•SDAcc can choose beneficial base classifiers with high probability.•SDAcc, DFTwo and AccRein are able to thin the ensemble the most.•SDAcc, DFTwo and AccRein keep an excellent testing accuracy.