TuneNSearch: a hybrid transfer learning and local search approach for solving vehicle routing problems

2025-03-16

AI Summary

This work addresses the poor generalization and weak adaptability of multi-task learning in diverse Vehicle Routing Problem (VRP) variants. We propose a two-stage framework integrating transfer learning and local search: first, pretraining a Transformer-enhanced residual edge graph attention network on high-complexity multi-depot VRPs to learn robust joint node-edge representations; second, multi-task fine-tuning to adapt to various VRP variants, augmented by lightweight, seamlessly embedded 2-opt/Or-opt local search for solution refinement. Key contributions include: (i) the first pretraining paradigm based on multi-depot VRPs; (ii) a novel residual edge graph attention mechanism; and (iii) end-to-end co-design of learned solving procedures with local search. Experiments demonstrate state-of-the-art performance across multiple VRP variants—surpassing most specialized SOTA models—with only 20% of their training time, strong cross-distribution and cross-scale generalization, and significant solution-quality gains from local search at negligible computational overhead.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

This paper introduces TuneNSearch, a hybrid transfer learning and local search approach for addressing different variants of vehicle routing problems (VRP). Recently, multi-task learning has gained much attention for solving VRP variants. However, this adaptability often compromises the performance of the models. To address this challenge, we first pre-train a reinforcement learning model on the multi-depot VRP, followed by a short fine-tuning phase to adapt it to different variants. By leveraging the complexity of the multi-depot VRP, the pre-trained model learns richer node representations and gains more transferable knowledge compared to models trained on simpler routing problems, such as the traveling salesman problem. TuneNSearch employs, in the first stage, a Transformer-based architecture, augmented with a residual edge-graph attention network to capture the impact of edge distances and residual connections between layers. This architecture allows for a more precise capture of graph-structured data, improving the encoding of VRP's features. After inference, our model is also coupled with a second stage composed of a local search algorithm, which yields substantial performance gains with minimal computational overhead added. Results show that TuneNSearch outperforms many existing state-of-the-art models trained for each VRP variant, requiring only one-fifth of the training epochs. Our approach demonstrates strong generalization, achieving high performance across different tasks, distributions and problem sizes, thus addressing a long-standing gap in the literature.

Problem

• Hybrid approach for solving vehicle routing problems
• Improves model adaptability without compromising performance
• Combines transfer learning with local search for efficiency

Innovation

• Hybrid transfer learning and local search approach
• Transformer-based architecture with residual edge-graph attention
• Pre-trained reinforcement learning model for multi-depot VRP