Abstract—To solve large-scale constraint satisfaction problems, CSPs, ant colony optimization, ACO, based meta-heuristics has been effective. Many methods based on ACO have been proposed including the cunning Ant System, cAS. However, some of these methods cannot be stable to solve CSPs. In this paper, we propose an ant colony optimization based meta-heuristics with multi pheromone trails. Artificial ants construct candidate assignments by referring several pheromone trail graphs to solve CSP instances. We also implement the proposed model to cAS method and demonstrate how our method is effective for solving large scale and hard graph coloring problems that are one of typical examples of CSPs.
Index Terms—Ant colony optimization, constraint satisfaction, graph coloring, meta heuristics.
The authors are with Takushoku University, Hachioji, Tokyo 193-0985 Japan (e-mail: phsl.masukane@gmail.com, mizuno@cs.takushoku-u.ac.jp).
Cite: Takuya Masukane and Kazunori Mizuno, "Solving Constraint Satisfaction Problems by Cunning Ants with multi-Pheromones," International Journal of Machine Learning and Computing vol. 8, no. 4, pp. 361-366, 2018.