Coevolutionary Algorithms for Realization of Intelligent Systems
Hyo-Byung Jun and Kwee-Bo Sim
Robotics and Intelligent Information System Laboratory School of Electrical and Electronic Engineering Chung-Ang University 221, Huksuk-Dong, Dongjak-Ku, Seou1156-756, Korea
Received:December 14, 1998Accepted:July 5, 1999Published:October 20, 1999
Keywords:Simple genetic algorithm, Schema theorem, Intelligent system, Co-evolutionary algorithm
The simple genetic algorithm (SGA) proposed by J. H. Holland uses population-based optimization based on Darwinian natural selection. The theoretical foundations of GA are the Schema Theorem and Building Block Hypothesis. Although GA does well in many applications in optimization, it does not guarantee the convergence to a global optimum in some problems. In designing intelligent systems, since there is no deterministic solution, heuristic trial-and error is usually used to determine system parameters. An alternative is a coevolutionary system, where 2 populations constantly interact and coevolve. We review coevolutionary algorithms and propose coevolutionary schemes designing intelligent systems based on the relationship between system components.
Cite this article as:H. Jun and K. Sim, “Coevolutionary Algorithms for Realization of Intelligent Systems,” J. Adv. Comput. Intell. Intell. Inform., Vol.3 No.5, pp. 418-426, 1999.Data files: