Clustering by pairwise similarity
dc.contributor.author | Li, Qilin | |
dc.contributor.supervisor | Assoc. Prof. Wan-Quan Liu | |
dc.contributor.supervisor | Assoc. Prof. Ling Li | |
dc.date.accessioned | 2017-01-30T10:15:26Z | |
dc.date.available | 2017-01-30T10:15:26Z | |
dc.date.created | 2016-06-30T01:19:22Z | |
dc.date.issued | 2016 | |
dc.identifier.uri | http://hdl.handle.net/20.500.11937/1939 | |
dc.description.abstract |
We proposed two novel clustering approaches, AFS and AFSSC, to address the problems in image clustering, semantic learning and manifold learning, respectively. By applying fuzzy membership function for data representation and similarity measure in AFS clustering, the pairwise affinity relationship is revealed more clearly than the commonly used Euclidean distance. In AFSSC method, we formulate AFS for similarity matrix learning and map the matrix to a weighted graph in which the clustering problem can be solved by graph cut theory. | |
dc.language | en | |
dc.publisher | Curtin University | |
dc.title | Clustering by pairwise similarity | |
dc.type | Thesis | |
dcterms.educationLevel | MPhil | |
curtin.department | Department of Computing | |
curtin.accessStatus | Open access |