Learning from design experience in an agent-based design system

A learning mechanism based on findings in cognitive science was added to an agent-based computational design system to determine if this mechanism would enable the system to learn from its experiences an transfer useful design knowledge to new problems. Learning and transfer were evaluated by examining how well knowledge learned while solving one problem could be applied to solve the same problem more effectively as well as how well this knowledge could be transferred to new design problems. An existing system, A-Design, was enhanced by giving it the ability to learn and store useful design knowledge so that this knowledge can be used in new design problems. Three electromechanical design problems were used to evaluate this new learning mechanism, and results indicate that this basic, cognitively based, learning mechanism is successful at transferring design knowledge to new problems with a few limitations. This knowledge transfer leads to a more effective design process.