An Optimization Algorithm Based on the Small-World Phenomenon

We propose an optimization algorithm, called the small-world algorithm (SWA), based on searching mechanisms in social networks. The SWA emphasizes local rather than global search to find the solutions for optimization problems. We investigate two encoding strategies, binary encoding and decimal encoding, in the SWA and test them by function optimization and 01 multidimensional knapsack problems. Simulation studies show that compared with the corresponding genetic algorithm (GA) the SWA is capable of preserving diversity in the set of candidate solutions, avoiding premature of the population, overcoming the “GA deceptive problem” and it converges faster. These properties suggest that the SWA is a useful method for solving complicated optimization problems. Keyword small-world phenomenon, optimization algorithm, function optimization, 0-1 multidimensional knapsack problem