J. Martikainen and S.J. Ovaska (Finland)
Evolutionary computation, traveling salesman problem, fuzzy logic, divide and conquer.
In this paper we present an efficient decomposition tech nique to speed up evolutionary algorithms when dealing with large scale optimization problems. Divide and conquer methods aim to solving problems in smaller entities and then combining the sub-solutions to form complete solu tions. Often the optimal way to divide the problem varies as the evolutionary algorithm proceeds, thus making a static decomposition not the best approach. In this paper, we pre sent a fuzzy controlled decomposition algorithm and show how it outperforms a traditional evolutionary algorithm.
Important Links:
Go Back
