Fuzzy-clustering methods, such as fuzzy k-means and Expectation Maximization, allow an object to be assigned to multiple clusters with different degrees of membership. However, the memberships that result from fuzzy-clustering algorithms are difficult to b
GeometricVisualizationofClustersObtainedfromFuzzy
ClusteringAlgorithms
LuisRueda andYuanquanZhang
Abstract
Fuzzy-clusteringmethods,suchasfuzzyk-meansandExpectationMaximization,allowanobjecttobeassignedtomultipleclusterswithdi erentdegreesofmembership.However,themembershipsthatresultfromfuzzy-clusteringalgorithmsaredi culttobeanalyzedandvisualized.Thememberships,usuallyconvertedto0-1values,arevisualizedusingparallelcoordinatesordi erentcolorshades.Inthispaper,weproposeanewapproachtovisualizefuzzy-clustereddata.Theschemeisbasedonageometricvisualization,andworksbygroupingtheobjectswithsimilarclustermembershipstowardstheverticesofahyper-tetrahedron.Theproposedmethodshowsclearadvantagesovertheexistingmethods,demonstratingitscapabilitiesforviewingandnavigatinginter-clusterrelationshipsinaspatialmanner.1Introduction
Manyclusteringalgorithmshavebeenproposedsofar,beingthemostimportantonesk-means
[21],fuzzyk-means[5],expectationmaximization(EM)[7],hierarchicalclustering[8],andself-organizingmaps[17].Someofthesemethodsgroupobjects,whichshareacertaindegreeofsimilarity,byassigningaprobabilityofmembershiptoeachcluster.Wecallthesemethods“fuzzy-clusteringalgorithms”.Inmolecularbiology,forexample,applyingfuzzy-clusteringtomicroarraydatabringstheadvantagethattheclusteringresultallowsagenetobeassignedtomorethanonecluster[5,9].Theproblem,however,ishowtoassigntheobjectstooneoftheclusters.Acommontechniquetodealwiththisistousea“cuto ”valueandassignanobjecttoacluster,ifitsmem-bershipprobabilityisabovethecuto .Ontheotherhand,visualizingthe“fuzzy”membershipofanobjectbelongingtodi erentclustersconstitutesinitselfanopenproblem.
Gaschetal.proposedamodi edhierarchicalvisualizationtoobservethefuzzy-clusteredgenesutilizingdi erentmembershipcuto values[10].Inthemembershiptable,eachobjectMemberoftheIEEE.DepartmentofComputerScience,UniversityofConcepcion,EdmundoLarenas215,Concepcion,VIIIRegion,Chile.Phone:+5641204305,Fax:+5641221770.E-mail:lrueda@inf.udec.cl. SchoolofComputerScience,UniversityofWindsor,401SunsetAve.,ON,N9B3P4,Canada.E-mail:kzhang@certicom.com.
A hybrid of fuzzy-link clustering and classification...algorithm is robust to noise and can discover ...visualization aspect, ignoring the geophysical ...
Using a non-hierarchical clustering algorithm and ...visualization and inspection of multivariate clusters...The synthetic catalogs (Fig.2b) were obtained by...
partitioning algorithm iteratively to produce a hierarchical tree of clusters. ...? Clustering helps to group / order information and is a visualization tool...
Anomaly Detection Using S Language Framework Clustering and Visualization of Intrusive..., pre-processing, algorithm development, data storage and visualization. ...
A Visualization Tool for... 暂无评价 77页 免费 A Clustering Algorithm f....Intelligence obtained by applying data mining to a database of French theses...
they provide effective visualization/exploration tools...(1998). Theoretrical Aspects of the SOM algorithm...: a contribution from fuzzy clustering methods. ...
approximation for large-scale terrain visualization ...It combines adaptive clustering based on quadtrees...geometric complexity and to approximate this non-...
The geometric objects are converted from their ...of multi-threshold and fuzzy clustering techniques...." Eurographics Workshop on Visualization in Scienti...
ective visualization and verbalization of unconscious...The information can be obtained from the answers ...A clustering algorithm for discrete objects is ...
Automatically Determining Number of Clusters Hua Yu...clustering algorithm, which gives visually appealing.... It's baiscally visualization of inter-document ...
热门文档