Loading...

New Hybrid Approaches Information Clustering Based on FCM Clustering and Optimization CFA

Farzin Far, Zohreh | 2016

1157 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 48290 (02)
  4. University: Sharif University of Technology
  5. Department: Mathematical Sciences
  6. Advisor(s): Ramezanian, Rasoul
  7. Abstract:
  8. Clustering algorithms are developed to provide general attitudes on database, recognizing latent structures and their more effective accessibility. Recently, broad studies are conducted on clustering since it is recognized as an important tool to explore and analyze data. Clustering is a fundamental learning operation without monitoring in data exploration which divides data into groups of objects so that objects in one group have the most similarity to each other and lowest similarity to objects in other groups. Some clustering algorithms such fuzzy clustering model (FCM) are widely used in clustering problem solution although this technique addresses local optimized searches. Additionally, in FCM, the preliminary locations of clusters are determined by user and the quality of this algorithm depends on the intensity of such preliminary location of clusters. As more complicated search method, Meta-heuristics can search response space more flexibility and more effectiveness. Fashion algorithm is a new and efficient meta-heuristics emanated from Fashion phenomenon in human communities. Each clustering algorithm categorizes data in different groups based on their varied abilities in clusters. To the same reason, by using different clustering techniques and their combinations, we expect to achieve better structures for clusters by utilizing the strengths of each clustering technique. In fact, the main aim of combined clustering is to look for the best clusters by using a combination of results of other algorithms. In contrary to traditional and classic clustering techniques that are less sustainable, combined clustering can generate more appropriate responses than other traditional clustering techniques in terms of preciseness, sustainability and velocity. In present dissertation, therefore, we attempt to combine FCM with Competitive Fashion algorithm to get rid of local summits and to achieve an overall optimized response automatically
  9. Keywords:
  10. Metaheuristic Method ; Optimization ; Clustering ; Fuzzy Clustering Method (FCM) ; Competitive Fation Algorithm

 Digital Object List

 Bookmark

...see more