Dynamic classifier selection for One-vs-One strategy: Avoiding non-competent classifiers
作者：
Highlights：
• Non-competent classifiers are one of the problems of One-vs-One strategy.
• Non-competence cannot be completely solved, but it can be reduced.
• A novel dynamic strategy is presented to reduce the non-competent classifiers.
• The neighbors of each instance are taken into account to avoid the non-competence.
• The new strategy outperforms the state-of-the-art aggregations.
摘要
Highlights•Non-competent classifiers are one of the problems of One-vs-One strategy.•Non-competence cannot be completely solved, but it can be reduced.•A novel dynamic strategy is presented to reduce the non-competent classifiers.•The neighbors of each instance are taken into account to avoid the non-competence.•The new strategy outperforms the state-of-the-art aggregations.
论文关键词：Multi-classification,Pairwise learning,One-vs-One,Decomposition strategies,Ensembles,Classifier selection
论文评审过程：Received 3 January 2012, Revised 29 November 2012, Accepted 30 April 2013, Available online 9 May 2013.
论文官网地址：https://doi.org/10.1016/j.patcog.2013.04.018
