| Abstract |
Nowadays, metaheuristics have become more and more important in solving the combinatorial optimization problems (COPs) for which the traditional methods (such as k-means, tabu search, and simulated annealing) are simply not powerful enough to obtain good solutions. Recently, spiral optimization (SO), a new metaheuristic that emulates the natural phenomena (such as swirl and low pressure) was proposed to solve the function optimization problem. In this thesis, we present an extension of SO, called a distributed spiral optimization (or DSO for short), for solving the COPs. The proposed algorithm differs from the original SO by (1) adding to the latter the k-means and oscillation operators to accelerate its convergence speed and (2) splitting the population into subpopulations so as to increase the diversity of the search, thus improving the quality of the clustering result. To evaluate the performance of the proposed algorithm, we compare it with the original SO and genetic k-means algorithm in solving the clustering problem. Moreover, to understand the impact of the parameter on the performance of DSO, we use Fourier amplitude sensitivity test (FAST) to analyze the dependency between the model and parameters. The results show that the proposed algorithm is quite promising. |
|---|
