Twin support vector machine based on adjustable large margin distribution for pattern classification
- PDF / 2,645,564 Bytes
- 19 Pages / 595.276 x 790.866 pts Page_size
- 57 Downloads / 169 Views
ORIGINAL ARTICLE
Twin support vector machine based on adjustable large margin distribution for pattern classification Liming Liu1 · Maoxiang Chu1 · Yonghui Yang1 · Rongfen Gong1 Received: 4 March 2019 / Accepted: 7 April 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract This paper researches the value of the margin distribution in binary classifier. The central idea of large margin distribution machine (LDM) is to optimize the margin distribution, such as maximizing the margin mean and minimizing the margin variance. Compared to support vector machine (SVM), LDM demonstrates the good generalization performance. In order to improve the generalization performance of twin support vector machine (TSVM), a twin support vector machine based on adjustable large margin distribution (ALD-TSVM) is proposed in this paper. Firstly, the margin distribution is redefined to construct a pair of adjustable supporting hyperplanes. Then, the redefined margin distribution is introduced onto TSVM to obtain the models of ALD-TSVM, including linear case and nonlinear case. ALD-TSVM is a general learning method which can be used in any place where TSVM and LDM can be applied. Finally, the novel method is compared with other classification algorithms by doing experiments on toy dataset, UCI datasets and image datasets. The experimental results show that ALD-TSVM obtains better classification performance. Keywords Pattern classification · Margin distribution · Twin support vector machine · Generalization performance
1 Introduction As a kind of novel machine learning method support vector machine (SVM) proposed by Vapnik et al. [1] is based on the statistical learning theory. It is built on the structural risk minimization and the margin maximization principles. And it finds an optimal classification hyperplane by solving a quadratic programming problem (QPP) [2]. SVM can solve the classification problems for small-scale samples and nonlinear separable samples. It has global optimality and generalization performance. It has been widely used in pattern classification and regression analysis [3–5]. Due to its superiority, many improved algorithms are derived, such as multicategory proximal SVM (MPSVM) [6], pinball loss SVM (Pin-SVM) [7], varying coefficient SVM (VCSVM) [8], hierarchical mixing linear SVMs (HMLSVMs) [9], fuzzy SVM (FSVM) [10], proximal SVM (PSTM) [11] and krein SVM (KSVM) [12]. * Maoxiang Chu [email protected] 1
School of Electronic and Information Engineering, University of Science and Technology Liaoning, Anshan 114051, Liaoning, China
Though SVM has many merits, it costs too much time to solve a large QPP. In order to reduce the time cost, Javedave et al. [13] proposed twin support vector machine (TSVM). TSVM is derived from generalized eigenvalue proximal SVM (GEPSVM) [14], and is similar to SVM in formula. It aims to find two nonparallel supporting hyperplanes by solving a pair of smaller sized QPPs and make each hyperplane be close to one class and be far from the other class as much a
Data Loading...
