[2N6-GS-1] Fundamental AI, theory: Constraint satisfaction and optimization

[2N6-GS-1-04] Dynamic Parameter Tuning with Swarm Intelligence for Constraint Satisfaction Problems

〇Takaaki Toya¹, Kazunori Mizuno¹, Takuya Masukane¹ (1. takushoku university)

Keywords:Constraint satisfaction, Swarm intelligence, Ant colony optimization

Ant colony optimization (ACO) has been an effective meta-heuristic to solve a constraint satisfaction problem (CSP).
There are some parameters in ACO. In ACO, the possibility of finding a solution depends on the balance of the values of parameters. The well-balanced parameters may raise the possibility of finding a solution in ACO. Setting the well-balanced parameters is difficult and time-consuming. We focus on PSOACO for automatic adjustment of the values of parameters. We apply PSOACO to a CSP. We improve PSOACO to raise the possibility of finding a solution for a CSP. We conduct the experiments to test the effectiveness of the improved PSOACO with graph coloring problems.