A Novel Twin Support Vector Machines with Manifold Regularization

Xiao Peng Hua; Xian Feng Li

doi:10.4028/www.scientific.net/AMM.644-650.1640

Paper Titles

A Statistics-Based Data Placement Strategy for Hybrid Storage
p.1620

Research on Geophysical Modeling Using Extreme Learning Machine for Scatterometer
p.1625

Research on Knowledge Resource Clustering Based on K-MEAN Algorithm
p.1629

Study on Model for Structural Reliability Based on Dynamic Interfering Theory
p.1633

A Novel Twin Support Vector Machines with Manifold Regularization
p.1640

The Special Form of Second Order Accuracy N-S Equations for Incompressible Fluid
p.1644

The Algebraic Algorithm for Solving the Path Planning Problem of a Weighted Directed Graph
p.1648

Approximation Analysis for a New Kind of Neural Network Using Rational Spline Weight Functions
p.1654

Complexity Analysis of Neural Network Based on Rational Spline Weight Functions
p.1658

HomeApplied Mechanics and MaterialsApplied Mechanics and Materials Vols. 644-650A Novel Twin Support Vector Machines with Manifold...

A Novel Twin Support Vector Machines with Manifold Regularization

Abstract:

Twin support vector machine (TWSVM), as a variant of the generalized eigenvalue proximal support vector machine (GEPSVM), attempts to improve the generalization of GEPSVM, whose solution follows from solving two quadratic programming problems (QPPs), each of which is smaller than in a standard SVM. Unfortunately, TWSVM fails to fully consider the local geometry structure and the local underlying descriminant information inside the samples that may be important for classification performance and only preserves the global data structure. In this paper, a novel TWSVM with manifold regularization is proposed by introducing the basic idea of the locality preserving within-class scatter matrix (LPWSM) into TWSVM. We termed this method manifold TWSVM (MTWSVM). MTWSVM not only retains the superior characteristics of TWSVM, but also preserves the local geometry structure between samples and shows the local underlying discriminant information. Experimental results confirm the effectiveness of our method.