Solving Global Optimization Problems Using MANGO
Traditional approaches for solving global optimization problems generally rely on a single algorithm. The algorithm may be hybrid or applied in parallel. Contrary to traditional approaches, this paper proposes to form teams of algorithms to tackle global optimization problems. Each algorithm is embodied and ran by a software agent. Agents exist in a multiagent system and communicate over our proposed MultiAgent ENvironment for Global Optimization (MANGO). Through communication and cooperation, the agents complement each other in tasks that they cannot do on their own. This paper gives a formal description of MANGO and outlines design principles for developing agents to execute on MANGO. Our case study shows the effectiveness of multiagent teams in solving global optimization problems.
|Keywords||Local Search Global Optimization Multiagent System Trust Region Global Optimization Problem|
|Persistent URL||dx.doi.org/10.1007/978-3-642-01665-3_79, hdl.handle.net/1765/117995|
|Journal||Lecture Notes in Computer Science|
Gunay, A., Oztoprak, F., Birbil, S.I., & Yolum, P. (2009). Solving Global Optimization Problems Using MANGO. In KES-AMSTA 2009: Agent and Multi-Agent Systems: Technologies and Applications (Vol. 5559, pp. 783–792). doi:10.1007/978-3-642-01665-3_79