An Adaptive Problem-Solving Solution to Large-Scale Scheduling Problems

Although the general class of most scheduling problems is NP-hard in worst-case complexity, in practice, domain-specific techniques frequently solve problems in much better than exponential time. Unfortunately, constructing special-purpose systems is a knowledge–intensive and time-consuming process that requires a deep understanding of the domain and problem-solving architecture. The goal of our work is to develop techniques to allow for automated learning of an effective domain-specific search strategy given a general problem solver with a flexible control architecture. In this approach, a learning system explores a space of possible heuristic methods a strategy well-suited to the regularities of the given domain and problem distribution. We discuss an application of our approach to scheduling satellite communications. Using problem distributions based on actual mission requirements, our approach identifies strategies that not only decrease the amount of CPU time required to produce schedules, but also increase the percentage of problems that are solvable within computational resource limitations.