k均值算法家族和子空间聚类.pdfVIP

计算机研究与发展 ISSN1000—1239／CN11—1777／TP of Researchand 42(增刊)：55～60，2005 Journal Computer Development 庀．均值算法家族和子空间聚类 黄哲学1 徐 军1 景丽萍1 高 阳2 叶允明3 李雄飞4 1(香港大学电子商务科技研究所香港) 2(南京大学计算机软件新技术国家重点实验室南京210093) 3(哈尔滨工业大学深圳研究生院深圳518055) 4(吉林大学计算机科学与技术学院长春130012) (jhuang@eti．hku．hk) A ofk-Means and Family TypeAlgorithmsSubspaceClustering Li HuangZhexuel，XuJunl，JingLipin91，GaoYan92，YeYunmin93，andXiongfei4 1(E-BusinessInstitute，Theof Technology UniversityHongKong，HongKong) 2(State Novel 210093) KeyLaboratoryforSoftwareTechnology，NanjingUniversity，Nanjing Graduate Institute 3(ShenzhenSchool，HarbinofTechnology，Shenzhen518055) Scienceand 4(DepartmentofComputer Technology，JilinUniversity，Changchun130012) 忌．meansisoneofthemost usedin the Abstract important algorithmsmanydisciplines．Inpast clustering of havebeen four variantsthe MacQueen’S愚一means decades，many original algorithm developed，which formak-meansof todate．Inthis the membersofthe走一means familyclusteringalgorithms paperkey their versionsarereviewedfirst．Then，the family(i．e．，忌．means，忌一modesand忌一prototypes)andfuzzy recent ofthenew忌一means withvariable andtheir in development typealgorithms weighting applications arediscussed．Thetextdocumentdataisusedtodemonstratethatthenewk—means subspaceclustering withvariable other愚一means intext clustering． algorithms