The Experts below are selected from a list of 145383 Experts worldwide ranked by ideXlab platform

Xinjun Peng - One of the best experts on this subject based on the ideXlab platform.

  • structural twin parametric margin Support Vector Machine for binary classification
    Knowledge Based Systems, 2013
    Co-Authors: Xinjun Peng, Yifei Wang
    Abstract:

    Twin parametric-margin Support Vector Machine (TPMSVM) determines the more flexible parametric-margin hyperplanes through a pair of quadratic programming problems (QPPs) compared with classical Support Vector Machine (SVM). However, it ignores the prior structural information in data. In this paper, we present a structural twin parametric-margin Support Vector Machine (STPMSVM) for classification. The two optimization problems of STPMSVM focus on the structural information of the corresponding classes based on the cluster granularity, which is vital for designing a good classifier in different real-world problems. Furthermore, two Mahalanobis distances are respectively introduced into its corresponding QPPs based on the structural information. STPMSVM has a special case of TPMSVM when each ellipsoid cluster is a unit ball in a reproducing kernel Hilbert space. Experimental results demonstrate that STPMSVM is often superior in generalization performance to other learning algorithms.

  • a twin hypersphere Support Vector Machine classifier and the fast learning algorithm
    Information Sciences, 2013
    Co-Authors: Xinjun Peng
    Abstract:

    This paper formulates a twin-hypersphere Support Vector Machine (THSVM) classifier for binary recognition. Similar to the twin Support Vector Machine (TWSVM) classifier, this THSVM determines two hyperspheres by solving two related Support Vector Machine (SVM)-type problems, each one is smaller than the classical SVM, which makes the THSVM be more efficient than the classical SVM. In addition, the THSVM avoids the matrix inversions in its two dual quadratic programming problems (QPPs) compared with the TWSVM. By considering the characteristics of the dual QPPs of THSVM, an efficient Gilbert's algorithm for the THSVM based on the reduced convex hull (RCH) instead of directly optimizing its pair of QPPs is further presented. Computational results on several synthetic as well as benchmark datasets indicate the significant advantages of the THSVM classifier in the computational time and test accuracy.

  • tpmsvm a novel twin parametric margin Support Vector Machine for pattern recognition
    Pattern Recognition, 2011
    Co-Authors: Xinjun Peng
    Abstract:

    A novel twin parametric-margin Support Vector Machine (TPMSVM) for classification is proposed in this paper. This TPMSVM, in the spirit of the twin Support Vector Machine (TWSVM), determines indirectly the separating hyperplane through a pair of nonparallel parametric-margin hyperplanes solved by two smaller sized Support Vector Machine (SVM)-type problems. Similar to the parametric-margin @[email protected]?Support Vector Machine ([email protected]@?SVM), this TPMSVM is suitable for many cases, especially when the data has heteroscedastic error structure, that is, the noise strongly depends on the input value. But there is an advantage in the learning speed compared with the [email protected]@?SVM. The experimental results on several artificial and benchmark datasets indicate that the TPMSVM not only obtains fast learning speed, but also shows good generalization.

Yuanhai Shao - One of the best experts on this subject based on the ideXlab platform.

  • weighted linear loss twin Support Vector Machine for large scale classification
    Knowledge Based Systems, 2015
    Co-Authors: Yuanhai Shao, Weijie Chen, Zhen Wang, Chunna Li, Naiyang Deng
    Abstract:

    In this paper, we formulate a twin-type Support Vector Machine for large-scale classification problems, called weighted linear loss twin Support Vector Machine (WLTSVM). By introducing the weighted linear loss, our WLTSVM only needs to solve simple linear equations with lower computational cost, and meanwhile, maintains the generalization ability. So, it is able to deal with large-scale problems efficiently without any extra external optimizers. The experimental results on several benchmark datasets indicate that, comparing to TWSVM, our WLTSVM has comparable classification accuracy but with less computational time.

  • multiple birth Support Vector Machine for multi class classification
    Neural Computing and Applications, 2013
    Co-Authors: Zhixia Yang, Yuanhai Shao, Xiangsun Zhang
    Abstract:

    For multi-class classification problem, a novel algorithm, called as multiple birth Support Vector Machine (MBSVM), is proposed, which can be considered as an extension of twin Support Vector Machine. Our MBSVM has been compared with the several typical Support Vector Machines. From theoretical point of view, it has been shown that its computational complexity is remarkably low, especially when the class number K is large. Based on our MBSVM, the dual problems of MBSVM are equivalent to symmetric mixed linear complementarity problems to which successive overrelaxation (SOR) can be directly applied. We establish our SOR algorithm for MBSVM. The SOR algorithm handles one data point at a time, so it can process large dataset that need no reside in memory. From practical point of view, its accuracy has been validated by the preliminary numerical experiments.

  • a regularization for the projection twin Support Vector Machine
    Knowledge Based Systems, 2013
    Co-Authors: Yuanhai Shao, Weijie Chen, Zhen Wang, Naiyang Deng
    Abstract:

    For the recently proposed projection twin Support Vector Machine (PTSVM) [1], we propose a simpler and reasonable variant from theoretical point of view, called projection twin Support Vector Machine with regularization term, RPTSVM for short. Note that only the empirical risk minimization is introduced in primal problems of PTSVM, incurring the possible singularity problems. Our RPTSVM reformulates the primal problems by adding a maximum margin regularization term, and, therefore, the singularity is avoided and the regularized risk principle is implemented. In addition, the nonlinear classification ignored in PTSVM is also considered in our RPTSVM. Further, a successive overrelaxation technique and a genetic algorithm are introduced to solve our optimization problems and to do the parameter selection, respectively. Computational comparisons of our RPTSVM against original PTSVM, TWSVM and MVSVM indicate that our RPTSVM obtains better generalization than others.

  • least squares recursive projection twin Support Vector Machine for classification
    Pattern Recognition, 2012
    Co-Authors: Yuanhai Shao, Naiyang Deng, Zhimin Yang
    Abstract:

    In this paper we formulate a least squares version of the recently proposed projection twin Support Vector Machine (PTSVM) for binary classification. This formulation leads to extremely simple and fast algorithm, called least squares projection twin Support Vector Machine (LSPTSVM) for generating binary classifiers. Different from PTSVM, we add a regularization term, ensuring the optimization problems in our LSPTSVM are positive definite and resulting better generalization ability. Instead of usually solving two dual problems, we solve two modified primal problems by solving two systems of linear equations whereas PTSVM need to solve two quadratic programming problems along with two systems of linear equations. Our experiments on publicly available datasets indicate that our LSPTSVM has comparable classification accuracy to that of PTSVM but with remarkably less computational time.

Qiu Xuena - One of the best experts on this subject based on the ideXlab platform.

  • face recognition method based on Support Vector Machine
    Computer Engineering, 2009
    Co-Authors: Qiu Xuena
    Abstract:

    This paper presents a face recognition method based on Binary Edge Map(BEM) and Support Vector Machine(SVM).It uses BEM as face representation and employs SVM as face classifiers.The BEM is extracted using Locally Adaptive Threshold(LAT) method.Experimental results show that a face recognition rate of 92.73% can be achieved on 165 Yale face images and 95.62% can be achieved on 798 AR images,indicating that the proposed approach is robust to varying illumination.

Sugen Chen - One of the best experts on this subject based on the ideXlab platform.

  • multiple birth least squares Support Vector Machine for multi class classification
    International Journal of Machine Learning and Cybernetics, 2017
    Co-Authors: Sugen Chen
    Abstract:

    Least squares twin Support Vector Machine (LSTSVM) was initially designed for binary classification. However, practical problems often require the discrimination more than two categories. To tackle multi-class classification problem, a novel algorithm, called multiple birth least squares Support Vector Machine (MBLSSVM), is proposed. Our MBLSSVM solves K quadratic programming problems (QPPs) to obtain K hyperplanes, each problem is similar to binary LSTSVM. Comparison against the Multi-LSTSVM, Multi-TWSVM, MBSVM and our MBLSSVM on both UCI datasets and ORL, YALE face datasets illustrates the effectiveness of the proposed method.

Shifei Ding - One of the best experts on this subject based on the ideXlab platform.

  • An improved multiple birth Support Vector Machine for pattern classification
    Neurocomputing, 2017
    Co-Authors: Xiekai Zhang, Shifei Ding
    Abstract:

    Multiple birth Support Vector Machine is a novel Machine learning algorithm for multi-class classification, which is considered as an extension of twin Support Vector Machine. Compared with training speeds of other multi-class classifiers based on twin Support Vector Machine, the training speed of multiple birth Support Vector Machine is faster, especially when the number of class is large. However, one of the disadvantages of multiple birth Support Vector Machine is that when used to deal with some datasets such as "Cross planes" datasets, multiple birth Support Vector Machine is likely to get bad results. In order to deal with this, we propose an improved multiple birth Support Vector Machine. We add a modified item into multiple birth Support Vector Machine to make the variance of the distances from each samples of a given class to their hyperplanes as small as possible. To predict a new sample, our method first determines an interval for each class depending on the distances between training samples and their hyperplanes, and then classifies the new sample depending on the distances between hyperplanes and the new sample which are in the corresponding intervals. In addition, smoothing technique is applied on our model, the first time it was used in multi-class twin Support Vector Machine. The experimental results on artificial datasets and UCI datasets show that the proposed algorithm is efficient and has good classification performance.