Robust support vector machine with bullet hole image classification
Authors:
- Wenfang XieWXWenfang Xie
- 1 November 2002
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man and Cybernetics, Part C (Applications and Reviews)
- Vol. 32 (4) , 440-448
Abstract
This paper proposes a robust support vector machine for pattern classification, which aims at solving the over-fitting problem when outliers exist in the training data set. During the robust training phase, the distance between each data point and the center of class is used to calculate the adaptive margin. The incorporation of the average techniques to the standard support vector machine (SVM) training makes the decision function less detoured by outliers, and controls the amount of regularization automatically. Experiments for the bullet hole classification problem show that the number of the support vectors is reduced, and the generalization performance is improved significantly compared to that of the standard SVM training.Keywords
This publication has 7 references indexed in Scilit:
- Using class-center vectors to build support vector machines2003
- Training support vector machines: an application to face detection2002
- Pedestrian detection using wavelet templates2002
- Bullet-hole image classification with support vector machines2000
- Adaptive margin support vector machines for classification1999
- A Tutorial on Support Vector Machines for Pattern RecognitionData Mining and Knowledge Discovery, 1998
- A training algorithm for optimal margin classifiers1992