Learning on Relational Data: Prototype-Based Classification of Attributed Graphs [Elektronische Ressource] / Shankar Deepak Srinivasan. Betreuer: Klaus Obermayer

Learning on Relational Data:Prototype-Based Classification ofAttributed Graphsvorgelegt vonM.Sc. (Engg.)Shankar Deepak Srinivasanaus Chennai, Indienvon der Fakult¨at IV - Elektrotechnik und Informatikder Technischen Universit¨at Berlinzur Erlangung des akademischen GradesDoktor der Naturwissenschaften- Dr. rer. nat. -Genehmigte DissertationTag der wissenschaftlichen Aussprache: 19. August 2011Promotionsausschuss:Vorsitzender: Prof. Dr. O. HellwichBerichter: Prof. Dr. K. ObermayerBerichterin: Prof. Dr. B. HammerBerlin, 2011D 832AcknowledgementsI thank my Doktorvater Prof. Klaus Obermayer for his advice andsupport. I wish to thank Brijnesh J. Jain for the many discussions,critical arguments and support over the last three years, JohannesMohr, Prof. Manfred Opper for their comments and insights, Mr.Francois-Xavier Dup´eand Prof. LucBrun forproviding theSHAPESdataset. I am grateful to members of the NI group both past andpresent for providing an atmosphere of support and enthusiasm. Spe-cial thanks to Max G¨ottner for running the experiments on SHAPESand COIL datasets. I would like to thank the efforts of the NISekretariat- Ms. Camilla Bruns and Ms. Gabriele R¨osler and BCCNadministration-especiallyMs. MargretFranke,Dr. VanessaCasagrandeand Dr. Daniela Pelz (formerly at BCCN Berlin) in helping me han-dle all manner of bureaucratic and official issues.