Towards Omni-generalizable Neural Methods for Vehicle Routing Problems
This addresses generalization challenges in VRPs for logistics and optimization applications, but it is incremental as it builds on existing meta-learning approaches.
The paper tackles the limited generalization of neural heuristics for vehicle routing problems (VRPs) by proposing a meta-learning framework that enables fast adaptation to new tasks with varying sizes and distributions, achieving effective performance on TSP and CVRP instances in experiments.
Learning heuristics for vehicle routing problems (VRPs) has gained much attention due to the less reliance on hand-crafted rules. However, existing methods are typically trained and tested on the same task with a fixed size and distribution (of nodes), and hence suffer from limited generalization performance. This paper studies a challenging yet realistic setting, which considers generalization across both size and distribution in VRPs. We propose a generic meta-learning framework, which enables effective training of an initialized model with the capability of fast adaptation to new tasks during inference. We further develop a simple yet efficient approximation method to reduce the training overhead. Extensive experiments on both synthetic and benchmark instances of the traveling salesman problem (TSP) and capacitated vehicle routing problem (CVRP) demonstrate the effectiveness of our method. The code is available at: https://github.com/RoyalSkye/Omni-VRP.