Paper Title
Ant System Combined with Autonomy and Cooperativeness

Abstract
Ant Colony Optimization (ACO) is well-known as an optimization tool based on Ant System (AS) inspired by real ants’ foraging behavior. In ACO, agents in ACO model solve an optimization problem like a traveling salesman problem (TSP) by behaving cooperatively based on information shared by all agents. However, the system of conventional ACO models has an issue for falling into a local solution because of the mechanical and heteronomous behavior of all agents. In this study, we construct a novel ACO model that achieved a balance between the system convergence and the divergence for improving an ability to find better solutions. We name this model ACAS (Ant System combined with Autonomy and Cooperativeness). ACAS is an extended version of AS adding two functions to AS. The behavior of agents in ACAS is in consideration of the fact that real ants realize a balance between convergence to a specific feeding ground and spreading to various feeding grounds. We applied ACAS to various datasets of symmetric TSP and found that ACAS performed better than AS. Index Terms - Ant Colony Optimization, Traveling Salesman Problem, Exploitation, Exploration