Neural Deconstruction Search for Vehicle Routing Problems
This addresses vehicle routing problems for logistics and optimization, presenting a novel paradigm shift from sequential construction.
The authors tackled vehicle routing problems by introducing an iterative search framework where a neural policy deconstructs solutions and collaborates with a greedy insertion algorithm to rebuild them, matching or surpassing state-of-the-art operations research methods across three challenging problems of various sizes.
Autoregressive construction approaches generate solutions to vehicle routing problems in a step-by-step fashion, leading to high-quality solutions that are nearing the performance achieved by handcrafted operations research techniques. In this work, we challenge the conventional paradigm of sequential solution construction and introduce an iterative search framework where solutions are instead deconstructed by a neural policy. Throughout the search, the neural policy collaborates with a simple greedy insertion algorithm to rebuild the deconstructed solutions. Our approach matches or surpasses the performance of state-of-the-art operations research methods across three challenging vehicle routing problems of various problem sizes.