![]() | Only 14 pages are availabe for public view |
Abstract The purpose of this thesis is to propose new algorithms for symbolic clustering.Till now the algorithms dealing with the symbolic objects are using the concepts of hierarchical techniques with using agglomerative methods or disaggregative methods as the core of the algorithm. The thesis oulines some of the major difficulties and drawbacks of two of the most popular techniques for symbolic clustering:Cluster/2 and Gowda and Diday algorithms.The proposed algorithms are inspired from efficient classical algorithms based on the concept of fuzziness and softness found in Fuzzy C-Means FCM and soft C Means SCM with using the advantage of iterating till some objectives function is satisfied.Extensive experiments on the proposed algorithms are presented in addition to an analysis of the computational complexities of all algorithms discussed. |