Structural twin parametric-margin support vector machine for binary classification

作者：

Highlights：

• A structural twin parametric-margin support vector machine (STPMSVM) classifier is presented.

• The structural information of corresponding classes based on cluster granularity is embedded into the optimization problems of STPMSVM.

• Two related Mahalanobis distances are respectively introduced into its corresponding QPPs based on structural information.

• STPMSVM degenerates into TPMSVM when each ellipsoid cluster is a unit ball in a reproducing kernel Hilbert space.

• STPMSVM is often superior in generalization performance to other learning algorithms.

摘要

•A structural twin parametric-margin support vector machine (STPMSVM) classifier is presented.•The structural information of corresponding classes based on cluster granularity is embedded into the optimization problems of STPMSVM.•Two related Mahalanobis distances are respectively introduced into its corresponding QPPs based on structural information.•STPMSVM degenerates into TPMSVM when each ellipsoid cluster is a unit ball in a reproducing kernel Hilbert space.•STPMSVM is often superior in generalization performance to other learning algorithms.

论文关键词：Binary classification,Support vector machine,Parametric margin,Structural granularity,Mahalanobis distance

论文评审过程：Received 1 December 2012, Revised 1 March 2013, Accepted 18 April 2013, Available online 25 April 2013.

论文官网地址：https://doi.org/10.1016/j.knosys.2013.04.013