Evolutionary-modified fuzzy nearest-neighbor rule for pattern classification

作者：

Highlights：

• Proposed a new fuzzy nearest neighbor rule with two new parameters.

• Proposed two new cost functions working in fuzzy class membership space.

• Verified the model reliability using experiments on several data sets.

• Statistically compared with several fuzzy based nearest neighbor rules.

• Implemented on graphical processing units for the sake of speed up.

摘要

•Proposed a new fuzzy nearest neighbor rule with two new parameters.•Proposed two new cost functions working in fuzzy class membership space.•Verified the model reliability using experiments on several data sets.•Statistically compared with several fuzzy based nearest neighbor rules.•Implemented on graphical processing units for the sake of speed up.

论文关键词：Pattern classification,Fuzzy nearest-neighbor rule,Multi-objective genetic algorithm,Graphical-processing unit

论文评审过程：Received 18 May 2016, Revised 10 July 2017, Accepted 11 July 2017, Available online 12 July 2017, Version of Record 14 July 2017.

论文官网地址：https://doi.org/10.1016/j.eswa.2017.07.013