International Journal of Applied Mathematics and Machine LearningAuthor's: Ming Hou and Liya Fan
Pages: [83] - [100]
Received Date: February 2, 2015
Submitted by: Jianqiang Gao.
DOI: http://dx.doi.org/10.18642/ijamml_7100121453
Motivated by nonparallel hyperplanes support vector machine (NHSVM), a
new regression method of data, named as nonparallel hyperplanes
support vector regression (NHSVR), is proposed in this paper. The
advantages of NHSVR have two aspects, one is considering the
minimization of structure risk by introducing a regularization term in
objective function, and another is finding two nonparallel hyperplanes
by solving a combined quadratic programming problem. In order to
verify the effectiveness of the propose method, a series of
comparative experiments are performed with TSVR, LTSVR, and 
on five evaluation indexes. The experiment
results show that the proposed NHSVR is an effective and efficient
algorithm for regression of data.
twin support vector regression, evaluation index, nonparallel hyperplanes, structural risk.
