Ensemble and constrained clustering with applications [Elektronische Ressource] / vorgelegt von Daniel Duarte Abdala

InformatikEnsemble and Constrained Clustering withApplicationsInaugural-Dissertationzur Erlangung des Doktorgradesder Naturwissenschaften im FachbereichMathematik und Informatikder Mathematisch-Naturwissenschaftlichen Fakult atder Westf alischen Wilhelms-Universit at Munstervorgelegt vonDaniel Duarte Abdalaaus S~ ao Paulo, Brazilien- 2010 -iiDekanin/Dekan: Prof. Dr. Matthias L oweProf. Dr. Xiaoyi JiangErster Gutachter:Zweiter Gutachter: Prof. Dr. Horst Bunke..............................Tag der mundlic hen Prufun g(en):Tag der Promotion:AbstractThe main focus of this thesis concerns the further developments in the areas of en-semble and constrained clustering. The goal of the proposed methods is to addressclustering problems, in which the optimal clustering method is unknown. Addi-tionally, by means of pairwise linkage constraints, it is possible to aggregate extrainformation to the clustering framework.Part I investigates the concept of ensemble clustering. It presents a comprehen-sive review of the state of the art in ensemble It follows by discussingthe impact of the ensemble variability in the nal consensual result. Visualizationof ensemble variability based on multidimensional scaling is also a topic addressedin this part. A software which is able to perform ensemble clustering using vari-ous existing consensus functions is also introduced.