Mathematical and Empirical Analysis of the Real World Tournament Selection

摘要

This paper provides a theoretical and empirical analysis of a recently proposed selection method, which is called 'Real World Tournament Selection' (RWTS). First of all, a characteristic of the selection probability of RWTS is analyzed. Secondly, the performance of RWTS is more widely tested at three benchmark problems - function optimization problems, the traveling salesman problem and the multdimensional knapsack problem. Finally, RWTS is compared with five well-known selection strategies. The experimental results show that it can achieve an encouraging performance, not only solution quality but also computation time.