000165281 001__ 165281
000165281 005__ 20190416220701.0
000165281 037__ $$aPOST_TALK
000165281 245__ $$aMethods for Clustering Multi-Layer Graphs in Mobile Networks
000165281 269__ $$a2011
000165281 260__ $$c2011
000165281 336__ $$aTalks
000165281 520__ $$aClustering on graphs has been studied extensively for years due to its numerous applications. However, in contrast to the classic problems, clustering in mobile and online social networks brings new challenges. In these scenarios, it is common that observational data contains multiple modalities of information reflecting different aspects of human behavior and social interactions. These interactions may be represented by a multi-layer graph that share the same set of vertices representing users, while having different layers representing different relationships among users. Intuitively, each graph should contribute to a better understanding of the underlying clusters from its own angle. It may be expected that a proper combination of the multiple graphs could lead to a better unified clustering of users' behavior and their social interactions. In this work we consider different methods to combine multi-layer graphs. In particular, we propose an efficient way to combine spectra of multiple graphs to form a “common spectrum”. To verify the suggested approach we tested it using mobile datasets. Also we compare the proposed approach with community detection methods based on modularity maximization over single and multiple layer graphs.
000165281 6531_ $$aMobile Social Network
000165281 6531_ $$aMulti-Layer Graph
000165281 6531_ $$aRegularization
000165281 6531_ $$aClustering
000165281 6531_ $$aLTS4
000165281 6531_ $$aLTS2
000165281 700__ $$0242933$$g193962$$aDong, Xiaowen
000165281 700__ $$0241061$$g101475$$aFrossard, Pascal
000165281 700__ $$0240428$$g120906$$aVandergheynst, Pierre
000165281 700__ $$aNefedov, Nikolai
000165281 7112_ $$dMay 31-June 1, 2011$$cMIT, Cambridge, Massachusetts, USA$$aInterdisciplinary Workshop on Information and Decision in Social Networks
000165281 8564_ $$uhttps://infoscience.epfl.ch/record/165281/files/MIT_WIDS_v1.pdf$$zn/a$$s32038$$yn/a
000165281 909C0 $$xU10380$$0252392$$pLTS2
000165281 909C0 $$pLTS4$$xU10851$$0252393
000165281 909CO $$qGLOBAL_SET$$pSTI$$ppresentation$$ooai:infoscience.tind.io:165281
000165281 917Z8 $$x193962
000165281 917Z8 $$x193962
000165281 917Z8 $$x193962
000165281 917Z8 $$x193962
000165281 917Z8 $$x193962
000165281 937__ $$aEPFL-TALK-165281
000165281 973__ $$sPUBLISHED$$aEPFL
000165281 980__ $$aTALK