%0 Journal Article %1 text.clustering.review.2004 %A Parsons, Lance %A Haque, Ehtesham %A Liu, Huan %D 2004 %J SIGKDD Exploration %K clustering review subspace text %N 1 %P 90-105 %T Subspace Clustering for High Dimensional Data: A Review %U http://www.sigkdd.org/explorations/issues/6-1-2004-06/parsons.pdf %V 6 %X Subspace clustering is an extension of traditional cluster- ing that seeks to ¯nd clusters in di®erent subspaces within a dataset. Often in high dimensional data, many dimen- sions are irrelevant and can mask existing clusters in noisy data. Feature selection removes irrelevant and redundant dimensions by analyzing the entire dataset. Subspace clus- tering algorithms localize the search for relevant dimensions allowing them to ¯nd clusters that exist in multiple, possi- bly overlapping subspaces. There are two major branches of subspace clustering based on their search strategy. Top- down algorithms ¯nd an initial clustering in the full set of dimensions and evaluate the subspaces of each cluster, it- eratively improving the results. Bottom-up approaches ¯nd dense regions in low dimensional spaces and combine them to form clusters. This paper presents a survey of the various subspace clustering algorithms along with a hierarchy orga- nizing the algorithms by their de¯ning characteristics. We then compare the two main approaches to subspace cluster- ing using empirical scalability and accuracy tests and discuss some potential applications where subspace clustering could be particularly useful.

@article{text.clustering.review.2004, abstract = {Subspace clustering is an extension of traditional cluster- ing that seeks to ¯nd clusters in di®erent subspaces within a dataset. Often in high dimensional data, many dimen- sions are irrelevant and can mask existing clusters in noisy data. Feature selection removes irrelevant and redundant dimensions by analyzing the entire dataset. Subspace clus- tering algorithms localize the search for relevant dimensions allowing them to ¯nd clusters that exist in multiple, possi- bly overlapping subspaces. There are two major branches of subspace clustering based on their search strategy. Top- down algorithms ¯nd an initial clustering in the full set of dimensions and evaluate the subspaces of each cluster, it- eratively improving the results. Bottom-up approaches ¯nd dense regions in low dimensional spaces and combine them to form clusters. This paper presents a survey of the various subspace clustering algorithms along with a hierarchy orga- nizing the algorithms by their de¯ning characteristics. We then compare the two main approaches to subspace cluster- ing using empirical scalability and accuracy tests and discuss some potential applications where subspace clustering could be particularly useful.}, added-at = {2009-03-07T02:17:59.000+0100}, author = {Parsons, Lance and Haque, Ehtesham and Liu, Huan}, biburl = {https://www.bibsonomy.org/bibtex/26e630a6a5b8175aa1b5f3f647b24a684/huiyangsfsu}, interhash = {fb6aa3c035c99b66778c30520585b73f}, intrahash = {6e630a6a5b8175aa1b5f3f647b24a684}, journal = {SIGKDD Exploration }, keywords = {clustering review subspace text}, number = 1, pages = {90-105}, timestamp = {2009-03-07T02:17:59.000+0100}, title = {Subspace Clustering for High Dimensional Data: A Review}, url = {http://www.sigkdd.org/explorations/issues/6-1-2004-06/parsons.pdf}, volume = 6, year = 2004 }