Artificial search agents with cognitive intelligence for binary optimization problems

Abstract Artificial intelligence techniques bring about new opportunities in problem solving. The notion such techniques have in common is learning mechanisms that are mostly problem and environment dependent. Although optimality is not guaranteed by these techniques, they draw attention due to being able to solve challenging optimization problems efficiently. Accordingly, the present study introduces a swarm-based optimization algorithm that is comprised of artificial search agents each with individual cognitive intelligence. In this technique, each agent is allowed to learn from problem space individually. Therefore, each of the search agents exhibits a different search characteristic. Nevertheless, they occasionally share information of the promising regions with each other. Thus, central swarm intelligence is also allowed to lead those independent search agents. Moreover, information-sharing techniques in the developed algorithm are designed as adaptive procedures so that search agents learn throughout generations by avoiding premature convergence and local optima problems as much as possible. The performance of the proposed algorithm is tested on a set of binary optimization problems including the set-union knapsack problem and the uncapacitated facility location problem, which have numerous real-life applications. All reported benchmarking problems are solved by the developed algorithm. As demonstrated by the comprehensive computational study and statistical tests, the proposed swarm-based algorithm significantly improves most of the published results.