| Summary: The Firefly algorithm is a population-based optimization algorithm. It has become popular in the field of optimization and has been applied to engineering practices. Recent works have failed to address how to find the global minimum, because their algorithm was trapped in the local minimum. Also, they were not able to provide a balance between exploration and exploitation. In this work, the Tidal Force formula has been applied to modify the Firefly algorithm. The proposed algorithm FAtidal brings a strategy into the optimization field. It is applied by using exploitation (Tidal Force) and keeping a balance between the exploration and exploitation on function suitability. Plate shaped, Steep Ridges, Unimodal and Multimodal Benchmark functions were used to compare experimental results. The study findings indicate that the Tidal Force Firefly algorithm outperforms the other existing modified Firefly algorithms. Another section of thesis proposes a strategy for clustering of the dataset with improved firefly algorithm. It is a procedure that partition data objects into the groups. Many algorithms could not overcome morphology, overlap and number of clusters problems at the same time. Clustering based on density is one of the best methods for those problems. This study proposed AFD algorithm based on Fuzzy and DBSCAN which works with the initialization of two parameters and FAtidal_DBSCAN algorithm proposed to reduce the sensitive paramters problems. In the experiments, It is demonstrated the proposed algorithms outperforms the other recently developed clustering algorithms.
Key Words: Firefly algorithm, Tidal force, Optimization, Swarm intelligence,
Global minimum, Clustering algorithm, Data mining, Fuzzy logic |
