Adaptive Agents at Work

Abstract

Evolutionary search methods enable a thorough allocation of recourses with respect to physical constraints, individual priorities and business objectives. If such an approach is used in a changing environment, adaptation is able to respond adequately to unforeseen events. In this way the planning of day-to-day logistics operation is done reactively. Similarly, a person-machine dialog can intervene into the resource planning in order to generate different plans, to explore scenarios or to modify the constraints. Due to the adaptive memory maintained by EAs, the search process may be suspended and reoriented, but it can be continued afterwards on the basis of the previously gathered knowledge. In this context we view an EA as a permanent, reactive, goal-oriented learning entity, or simply as an adaptive agent.