Near parameter free ant colony optimisation

Abstract

Ant colony optimisation, like all other meta-heuristic search processes, requires a set of parameters in order to solve combinatorial problems. These parameters are often tuned by hand by the researcher to a set that seems to work well for the problem under study or a standard set from the literature. However, it is possible to integrate a parameter search process within the running of the meta-heuristic without incurring an undue computational overhead. In this paper, ant colony optimisation is used to evolve suitable parameter values (using its own optimisation processes) while it is solving combinatorial problems. The results reveal for the travelling salesman and quadratic assignment problems that the use of the augmented solver generally performs well against one that uses a standard set of parameter values. This is attributed to the fact that parameter values suitable for the particular problem instance can be automatically derived and varied throughout the search process. © 2004 Springer-Verlag.