Off-line vs. On-line Tuning: A Study on MAX–MIN Ant System for the TSP

Stochastic local search algorithms require finding an appropriate setting of their parameters in order to reach high performance. The parameter tuning approaches that have been proposed in the literature for this task can be classified into two families: on-line and off-line tuning. In this paper, we compare the results we achieved with these two approaches. In particular, we report the results of an experimental study based on a prominent ant colony optimization algorithm, MAX– MIN Ant System, for the traveling salesman problem. We observe the performance of on-line parameter tuning for different parameter adaptation schemes and for different numbers of parameters to be tuned. Our results indicate that, under the experimental conditions chosen here, offline tuned parameter settings are preferable.