Intelligent Engineering Systems through Artificial Neural Networks
28 A Multi-Agent Organizational Model for Coevolutionary Optimization
Download citation file:
- Ris (Zotero)
- Reference Manager
This paper introduces MAS4EVO, a new multi-agent organizational model supporting the design and hybridization of evolutionary algorithms (EAs) including co-evolutionary genetic algorithms (CGAs). Representing EAs as an organizational multi-agent system allows dealing with simplicity, autonomy, modularity and robustness. It permits to explicit their characteristics in terms of topology, interaction and adaptation. MAS4EVO provides a complete framework for designing EAs, including a dedicated organizational model based on four specifications (structural, functional, dialogic and normative), allowing to fully describing EAs characteristics. MAS4EVO's utilization is demonstrated by the design of a cooperative and a competitive CGA. Their application using MAS4EVO's implementation, called DAFO (Distributed Agent Framework for Optimization) is illustrated on an emergent topology control problem in mobile hybrid ad hoc networks called injection network problem.