A Novel Metric for Comparing the Intelligence of Two Swarm Multiagent Systems

摘要

Background: Swarm systems are used for many difficult, even NP-hard problems solving. In the scientific literature, the swarm systems intelligence is mostly considered at the swarm level and it is based on some considerations, like autonomous learning, self-adaptation, evolution and efficient and flexible problems solving. It were identified as a necessity the elaboration of metrics that allow an effective and accurate measuring of swarm systems intelligence, taking also into consideration the variability of intelligence in the systems responses to different situations. Materials and Methods: In this study a novel metric, called MetrInt have been proposed for accurate comparison of two swarm systems intelligence. For proving the effectiveness of the metric, it was realized a case study for two swarm systems that mimic the biological ant colony, which solve a computational NP-hard problem, the traveling salesman problem. Results: The main result reported in this study consists in the proposed metric for comparison of two swarm systems intelligence. By applying the proposed metric, MetrInt, in the case study, it resulted that the two studied swarm multiagent systems can be considered as having the same intelligence level even there is a measurable numerical difference in intelligence. Based on this consideration, it can be concluded that both of them can be included in the same intelligence class. Conclusion: The proposed metric is appropriate for comparing two swarm systems with the same type of intelligence based on their intelligence level. It is also appropriate in the case of the design of swarm systems that should intelligently solve problems. The proposed metric is appropriate for comparison of two swarm systems intelligence. A swarm system could behave in different situations with lower or higher intelligence. MetrInt is an accurate metric, which takes into consideration the intelligence manifestation?s variability. It is effective even in the case of small differences in the intelligence of the swarm systems. Two swarm systems with the same intelligence level could be considered that belong to the same intelligence class.