Multi-sphere Support Vector Data Description for Outliers Detection on Multi-distribution Data

Xiao, Yanshan; Liu, Bo; Cao, Longbing; Wu, Xindong; Zhang, Chengqi; Zhang, Hao; Yang, Fengzhao; Cao, Jie

doi:10.1109/icdmw.2009.87

Cited by 33 publications

(12 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Mostly related to ours are the works of [17,18,25], which are shared with us the idea of using a set of hyperspheres as the domain of novelty. In [25], the procedure to discover a set of hyperspheres is ad hoc, heuristic and does not conform to any learning principle. The works of [17,18] are driven by the principle learning with minimum volume.…”

Section: Related Workmentioning

confidence: 95%

Mixture of hyperspheres for novelty detection

Nguyen¹,

Lai²,

Nguyen³

et al. 2016

Vietnam J Comput Sci

View full text Add to dashboard Cite

In this paper, we present a mixture of support vector data descriptions (mSVDD) for one-class classification or novelty detection. A mixture of optimal hyperspheres is automatically discovered to characterize data. The model includes two parts: log likelihood to control the fit of data to model (i.e., empirical risk) and regularization quantizer to control the generalization ability of model (i.e., general risk). Expectation maximization (EM) principle is employed to train our proposed mSVDD. We demonstrate the advantage of the proposed model: if learning mSVDD in the input space, it simulates learning a single hypersphere in the feature space and the accuracy is thus comparable, but the training time is significantly shorter.

show abstract

Section: Related Workmentioning

confidence: 95%

Mixture of hyperspheres for novelty detection

Nguyen¹,

Lai²,

Nguyen³

et al. 2016

Vietnam J Comput Sci

View full text Add to dashboard Cite

show abstract

“…It should be notified that, when the nonlinear solution is reduced to a set of linear solutions, the kernel trick becomes unnecessary, and it enables our framework to potentially scale up to large scale data sets. Recently, multisphere support vector data description (SVDD) methods [14], [15], [16] have also been introduced to extend the traditional one-class SVM. Our method differs from the Multi-sphere SVDD approach in that we employ the low-rank constraint to group sample set into different clusters automatically, and we also present two different optimization algorithms to address this problem.…”

Section: Outliersmentioning

confidence: 99%

Locality linear fitting one-class SVM with low-rank constraints for outlier detection

Shao

2014

2014 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

We propose a novel outlier detection approach in this paper, which learns the most accurate hyperspheres for the normal data through a top-down procedure. Conventional one-class support vector machine (SVM) based approaches aim to find nonlinear global solutions for all the normal data, with the benefit of kernel trick. However, those methods are intractable when data are in large-scale and inaccurate when data are under complex distributions. It's observed that high dimensional data, e.g., features of texts or images, are always sparse, and linear classifier usually performs well. A specific class of data seldom lie in one single subspace. In this paper, we propose to learn multiple discriminative hyper-spheres locally based on the data distributions, and fit them globally to formulate a more discriminative boundary for the normal data. By far, neural mechanisms used by human brain-mind for outlier detection are not known, however, the top-down strategy proposed in this paper would inspire understanding of the human neural mechanisms. The benefits of our model are twofolds. First, the distribution of each local cluster is much simpler than that in a global view, which makes the fitting processing for each individual cluster much easier and insensitive to the choice of kernel. In particular, we adopt low-rank constraints to find multiple clusters automatically. Secondly, the proposed approach trains the model linearly which tackles the largescale problem, substantially reducing training time and memory space. Extensive experimental results on three image databases demonstrate that our approach outperforms several related methods.

show abstract

“…In [19], authors propose to split the target class into sub-groups; each one is bounded by an SVDD. An other approach based on binary classification is proposed in [15]; where both the target class and the outliers are enclosed in two different hyperspheres.…”

Section: Support Vector Data Description (Svdd)mentioning

confidence: 99%

Enhancing the SVDD Accuracy in Intrusion Detection Systems by Removing External Voids

Kenaza

Bennaceur

Labed³

et al. 2016

2016 IEEE Trustcom/BigDataSE/Ispa

View full text Add to dashboard Cite

Abstract-This work aims to improve the accuracy of the SVDD-based Intrusion Detection Systems. In this study we are interested by approaches using only one-class classification, namely the class of normal user sessions. Sessions are modeled by vectors of points in a finite features space. The goal of using the SVDD in anomaly detection is to find the hypersphere with a minimal volume that encloses the entire scatter of points (i.e. the normal sessions). This paper discusses the general case where the shape of the scatter is arbitrary. In this case some voids can occur between the scatter and the boundary of the hypersphere, and mainly cause a distortion of the data description that reduces the accuracy of the detection. The objective of this work is to study and highlight the best techniques that help removing voids and thus improving the accuracy of the SVDD. Experimental results show that choosing the appropriate techniques and parameters can significantly improve the accuracy of the SVDD.

show abstract

Multi-sphere Support Vector Data Description for Outliers Detection on Multi-distribution Data

Cited by 33 publications

References 7 publications

Mixture of hyperspheres for novelty detection

Mixture of hyperspheres for novelty detection

Locality linear fitting one-class SVM with low-rank constraints for outlier detection

Enhancing the SVDD Accuracy in Intrusion Detection Systems by Removing External Voids

Contact Info

Product

Resources

About