Infinite Bayesian one-class support vector machine based on Dirichlet process mixture clustering
作者：
Highlights：
• We develop a novel OCC method based on the DPM clustering and the modified OCSVMs.
• Our method combines the reconstruction-based method and the boundary-based method.
• The reconstruction and bounding construction are jointly optimized in a Bayesian frame.
• Our method can improve the robustnessas well as the accuracy of classification performance.
• Experimental results show our method performs better than other related methods.
摘要
•We develop a novel OCC method based on the DPM clustering and the modified OCSVMs.•Our method combines the reconstruction-based method and the boundary-based method.•The reconstruction and bounding construction are jointly optimized in a Bayesian frame.•Our method can improve the robustnessas well as the accuracy of classification performance.•Experimental results show our method performs better than other related methods.
论文关键词：Dirichlet process mixture,One-class classifiers,One-class support vector machine,Gibbs sampling
论文评审过程：Received 1 August 2016, Revised 11 September 2017, Accepted 7 January 2018, Available online 11 January 2018, Version of Record 30 January 2018.
论文官网地址：https://doi.org/10.1016/j.patcog.2018.01.006
