Automated configuration of parallel machine dispatching rules by machine learning

Finding optimized machine schedules is a very important task that arises in many areas of industrial manufacturing. In practice dispatching rules are often used to create schedules within short run times. This paper investigates a method for automated specification of parameters for weighted dispatching rules. We define this problem as a machine learning task and propose a novel set of features to characterize problem instances. Experimental results with a practical dispatching rule show that our approach can obtain high quality solutions in short run times.