Scaling up solutions to storage location assignment problems by genetic programming

The Storage Location Assignment Problem (SLAP) is to find an optimal stock arrangement in a warehouse. This study presents a scalable method for solving large-scale SLAPs utilizing Genetic Programming (GP) and two sampling strategies. Given a large scale problem, a sub-problem is sampled from the original problem for our GP method to learn an allocation rule (in the form of a matching function). Then this rule can be applied to the original problem to generate solutions. By this approach, the allocation rule can be obtained in a much shorter time. When sampling the problem, the representativeness is a key factor that can largely affect the generalizability of the trained allocation rule. To investigate the effect of representativeness, two sampling strategies, namely the random sampling and filtered sampling, are proposed and compared in this paper. The filtered sampling strategy adopts more information about the problem structure to increase the similarity of the sampled problem and the entire problem. The results show that the filtered sampling performs significantly better than the random sampling in terms of both solution quality and success rate (i.e., the probability of generating feasible solutions for the large problem). The good performance of filtered strategy indicates the importance of sample representativeness on the scalability of the GP generated rules.