Application of Swarm Intelligence in Arbitrary Shaped Clustering

Please enable javascript in your browser.

Gharehyazie, Mohammad | 2010

977 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 40353 (19)
University: Sharif University of Technology
Department: Computer Engineering
Advisor(s): Bagheri Shouraki, Saeed
Abstract:
Clustering has great applications in various fields such as marketing, health, insurance, bioinformatics and etc. The assumption of regular parametric clusters is a common problem in current popular methods. This assumption is not valid in most real applications and greatly increases the clustering errors even to an unacceptable rate. Inspired by sardine fish, in this thesis we propose a new model with high elasticity factor that can cluster data without cluster shape constraints. This method uses the sardine fish model to augment clustering space dimension in order to achieve greater separability. The proposed method had some problems that were fixed and the final algorithm was finalized. The results show that the algorithm is quite successful in increasing the clustering space separability
Keywords:
Clustering ; Self-Organizing Map (SOM) ; Swarm Intelligence ; Arbitraryshaped Clusters