Publication Type

Conference Proceeding Article

Version

publishedVersion

Publication Date

7-2025

Abstract

Solving various types of vehicle routing problems (VRPs) using a unified neural solver has garnered significant attentions in recent years. Despite their effectiveness, existing neural multi-task solvers often fail to account for the geometric structures inherent in different tasks, which may result in suboptimal performance. To address this limitation, we propose a curvature-aware pre-training framework. Specifically, we leverage mixed-curvature spaces during the feature fusion stage, encouraging the model to capture the underlying geometric properties of each instance. Through extensive experiments, we evaluate the proposed pre-training strategy on existing neural multi-task solvers across a variety of testing scenarios. The results demonstrate that the curvature-aware pre-training approach not only enhances the generalization capabilities of existing neural VRP solvers on synthetic datasets but also improves solution quality on real-world benchmarks.

Keywords

vehicle routing problems, multi-task learning, curvature-aware pre-training, geometric deep learning, mixed-curvature spaces, feature fusion, neural solvers, generalization, benchmark evaluation, transportation optimization

Discipline

Artificial Intelligence and Robotics

Research Areas

Intelligent Systems and Optimization

Publication

Proceedings of the 42nd International Conference on Machine Learning

First Page

38066

Last Page

38101

City or Country

Vancouver, Canada

Additional URL

https://proceedings.mlr.press/v267/liu25b.html

Download

Share

COinS