Improving Search Efficiency Adopting Hill-Climbing to Ant Colony Optimization for Constraint Satisfaction Problems

To efficiently solve large-scale constraint satisfaction problems, CSPs, we propose an ant colony optimization based meta-heuristics combined with the hill-climbing approach. In our method, in order to improve search inefficiency which happens due to slow reconstruction of assignments of values to variables in the naive ant system, AS, min-conflict hill-climbing is applied to some assignments constructed ones by AS. This method is applied to large-scale and hard binary CSP instances in phase transition regions, whose experimental simulations demonstrate that our method is more efficient than AS.