Ryoji Tanabe and Tad Gonsalves
Interactive Evolutionary Computation, Ant Colony Optimization, Simulated Breeding
Interactive Evolutionary Computation (IEC) deals with the optimization of an objective function that is subjectively evaluated by repeated interaction with the user. The critical problem in IEC application areas is the load on the user because the iterative interaction between the human user and the system to evaluate the solutions produces stress and fatigue in the human user. Ant Colony Optimization (ACO) is a well-known Evolutionary Computation Algorithm. In this paper, we propose the Interactive Ant Colony Optimization (IACO), which is an extension of the ACO algorithm to the IEC application fields. The ACO algorithm has a parameter called evaporation rate which controls the speed of search convergence. If this parameter is tailored to the application problem, the IACO algorithms are able to accelerate the search and get the admissible solution. Owing to this control parameter, IACO has the potential of being a promising IEC algorithm. Our new algorithm called Simulated Breeding-based Interactive Ant System (SIAS) is fitted with Simulated Breeding as the evaluation-input method. To evaluate the performance of the proposed algorithm, we define the pseudo-IEC user based on the wellknown fitness landscape NK model and use it for the IEC simulation. The experimental results show that the performance of SIAS is superior to that of Interactive Genetic Algorithm (IGA) which is a typical IEC algorithm.
Important Links:
Go Back
