Managing information complexity of supply chains via agent-based genetic programming

This paper proposes agent-based formulation of a supply chain management (SCM) system for manufacturing firms. We model each firm as a decision-making agent, which communicates each other through the blackboard architecture in distributed artificial intelligence. To overcome the issues of conventional SCM systems, we employ the concept of information entropy, which represents the complexity of the purchase, sales, and inventory activities of each firm. Based on the idea, we implement an agent-based simulator to learn "good" decisions via genetic programming in a logic-programming environment. From intensive experiments, our simulator has shown good performance against the dynamic environmental changes.