Proximity multi-sphere support vector clustering

    Research output: Contribution to journalArticle

    8 Citations (Scopus)

    Abstract

    Support vector data description constructs an optimal hypersphere in feature space as a description of a data set. This hypersphere when mapped back to input space becomes a set of contours, and support vector clustering (SVC) employs these contours as cluster boundaries to detect clusters in the data set. However real-world data sets may have some distinctive distributions and hence a single hypersphere cannot be the best description. As a result, the set of contours in input space does not always detect all clusters in the data set. Another issue in SVC is that in some cases, it cannot preserve proximity notation which is crucial for cluster analysis, that is, two data points that are close to each other can be assigned to different clusters using cluster labelling method of SVC. To overcome these drawbacks, we propose Proximity Multi-sphere Support Vector Clustering which employs a set of hyperspheres to provide a better data description for data sets having distinctive distributions and a proximity graph to favour the proximity notation. Experimental results on different data sets are presented to evaluate the proposed clustering technique and compare it with SVC and other clustering techniques
    Original languageEnglish
    Pages (from-to)1309-1319
    Number of pages11
    JournalNeural Computing and Applications
    Volume22
    Issue number7-8
    DOIs
    Publication statusPublished - 2013

    Fingerprint

    Data description
    Cluster analysis
    Labeling

    Cite this

    @article{9d887b7017164580a2fe3fcf695a7bf2,
    title = "Proximity multi-sphere support vector clustering",
    abstract = "Support vector data description constructs an optimal hypersphere in feature space as a description of a data set. This hypersphere when mapped back to input space becomes a set of contours, and support vector clustering (SVC) employs these contours as cluster boundaries to detect clusters in the data set. However real-world data sets may have some distinctive distributions and hence a single hypersphere cannot be the best description. As a result, the set of contours in input space does not always detect all clusters in the data set. Another issue in SVC is that in some cases, it cannot preserve proximity notation which is crucial for cluster analysis, that is, two data points that are close to each other can be assigned to different clusters using cluster labelling method of SVC. To overcome these drawbacks, we propose Proximity Multi-sphere Support Vector Clustering which employs a set of hyperspheres to provide a better data description for data sets having distinctive distributions and a proximity graph to favour the proximity notation. Experimental results on different data sets are presented to evaluate the proposed clustering technique and compare it with SVC and other clustering techniques",
    keywords = "Clustering, Multi-sphere support vector clustering, Multi-sphere support vector data description, Proximity graph, Support vector clustering, Support vector data description",
    author = "Dat TRAN and Wanli MA and Dharmendra SHARMA",
    year = "2013",
    doi = "10.1007/s00521-012-1001-7",
    language = "English",
    volume = "22",
    pages = "1309--1319",
    journal = "Proximity multi-sphere support vector clustering",
    issn = "0941-0643",
    publisher = "Springer",
    number = "7-8",

    }

    Proximity multi-sphere support vector clustering. / TRAN, Dat; MA, Wanli; SHARMA, Dharmendra.

    In: Neural Computing and Applications, Vol. 22, No. 7-8, 2013, p. 1309-1319.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - Proximity multi-sphere support vector clustering

    AU - TRAN, Dat

    AU - MA, Wanli

    AU - SHARMA, Dharmendra

    PY - 2013

    Y1 - 2013

    N2 - Support vector data description constructs an optimal hypersphere in feature space as a description of a data set. This hypersphere when mapped back to input space becomes a set of contours, and support vector clustering (SVC) employs these contours as cluster boundaries to detect clusters in the data set. However real-world data sets may have some distinctive distributions and hence a single hypersphere cannot be the best description. As a result, the set of contours in input space does not always detect all clusters in the data set. Another issue in SVC is that in some cases, it cannot preserve proximity notation which is crucial for cluster analysis, that is, two data points that are close to each other can be assigned to different clusters using cluster labelling method of SVC. To overcome these drawbacks, we propose Proximity Multi-sphere Support Vector Clustering which employs a set of hyperspheres to provide a better data description for data sets having distinctive distributions and a proximity graph to favour the proximity notation. Experimental results on different data sets are presented to evaluate the proposed clustering technique and compare it with SVC and other clustering techniques

    AB - Support vector data description constructs an optimal hypersphere in feature space as a description of a data set. This hypersphere when mapped back to input space becomes a set of contours, and support vector clustering (SVC) employs these contours as cluster boundaries to detect clusters in the data set. However real-world data sets may have some distinctive distributions and hence a single hypersphere cannot be the best description. As a result, the set of contours in input space does not always detect all clusters in the data set. Another issue in SVC is that in some cases, it cannot preserve proximity notation which is crucial for cluster analysis, that is, two data points that are close to each other can be assigned to different clusters using cluster labelling method of SVC. To overcome these drawbacks, we propose Proximity Multi-sphere Support Vector Clustering which employs a set of hyperspheres to provide a better data description for data sets having distinctive distributions and a proximity graph to favour the proximity notation. Experimental results on different data sets are presented to evaluate the proposed clustering technique and compare it with SVC and other clustering techniques

    KW - Clustering

    KW - Multi-sphere support vector clustering

    KW - Multi-sphere support vector data description

    KW - Proximity graph

    KW - Support vector clustering

    KW - Support vector data description

    U2 - 10.1007/s00521-012-1001-7

    DO - 10.1007/s00521-012-1001-7

    M3 - Article

    VL - 22

    SP - 1309

    EP - 1319

    JO - Proximity multi-sphere support vector clustering

    JF - Proximity multi-sphere support vector clustering

    SN - 0941-0643

    IS - 7-8

    ER -