Support Vector Data Description Of The Application Of The Opposition Point Detection

The substance of novelty detection is a problem of one-class classification, novelty data can provide more information and more important information than the normal data. The technique of novelty detection is useful in applications such as fault diagnosis, hand written digit recognition, network abnormal intrusion detection, speaker recognition and several others.Support vector data description (SVDD) is a one-class classification method that developed from the statistic learning theory and support vector machine. The basic idea of SVDD method is to find a super-sphere in feature space and limit the volume of the sphere to be the smallest, in the same time include as possible as more target data and exclude as possible as more non-target sample data, thus the data can be classified. The paper researched briefly on SVDD and the method was applied to fault diagnosis.The paper's contents include: Basic learning theory, kernel technology, support vector machine(SVM) and SVM's characteristics; The theory of SVDD; the performance of common kernel functions on SVDD was researched, the gauss-kernel that possessed certain practicability was verified; The influence on SVDD probed by gauss-kernel-parameter, and experiments showed: SVDD method was adaptive to one-class classification of small-scale sample; According to distance similarity to reduce the training samples between support vectors locating on the classification boundary, the new-style method of SVDD-Similariry Reduction was proposed, and the method in terms of small-scale data had some effectiveness; Fault diagnosis employed SVDD to classify and used the wavelet packet decomposing technology to feature extraction, finally, good classification effect was obtained in experiments.