Deep policy dynamic programming for vehicle routing problems
From MaRDI portal
Publication:2170197
DOI10.1007/978-3-031-08011-1_14zbMATH Open1504.90175arXiv2102.11756OpenAlexW3132134635MaRDI QIDQ2170197FDOQ2170197
Authors: Wouter Kool, Herke van Hoof, Max Welling, Joaquim A. S. Gromicho 
Publication date: 30 August 2022
Abstract: Routing problems are a class of combinatorial problems with many practical applications. Recently, end-to-end deep learning methods have been proposed to learn approximate solution heuristics for such problems. In contrast, classical dynamic programming (DP) algorithms guarantee optimal solutions, but scale badly with the problem size. We propose Deep Policy Dynamic Programming (DPDP), which aims to combine the strengths of learned neural heuristics with those of DP algorithms. DPDP prioritizes and restricts the DP state space using a policy derived from a deep neural network, which is trained to predict edges from example solutions. We evaluate our framework on the travelling salesman problem (TSP), the vehicle routing problem (VRP) and TSP with time windows (TSPTW) and show that the neural policy improves the performance of (restricted) DP algorithms, making them competitive to strong alternatives such as LKH, while also outperforming most other 'neural approaches' for solving TSPs, VRPs and TSPTWs with 100 nodes.
Full work available at URL: https://arxiv.org/abs/2102.11756
Recommendations
- Learning the travelling salesperson problem requires rethinking generalization
- Learning heuristics for the TSP by policy gradient
- Solving the traveling salesperson problem with precedence constraints by deep reinforcement learning
- A restricted dynamic programming heuristic algorithm for the time dependent traveling salesman problem
- TLHSA and SACA: two heuristic algorithms for two variant VRP models
Artificial neural networks and deep learning (68T07) Programming involving graphs or networks (90C35) Combinatorial optimization (90C27) Dynamic programming (90C39)
Cites Work
- A note on two problems in connexion with graphs
- A Dynamic Programming Approach to Sequencing Problems
- An Optimal Algorithm for the Traveling Salesman Problem with Time Windows
- On the Theory of Dynamic Programming
- The vehicle routing problem: An overview of exact and approximate algorithms
- An approximate dynamic programming approach for the vehicle routing problem with stochastic demands
- Record breaking optimization results using the ruin and recreate principle
- Dynamic Programming Treatment of the Travelling Salesman Problem
- Dynamic Programming Strategies for the Traveling Salesman Problem with Time Window and Precedence Constraints
- A Hybrid Genetic Algorithm for Multidepot and Periodic Vehicle Routing Problems
- Vehicle Routing
- Restricted dynamic programming: a flexible framework for solving realistic VRPs
- Solving the job-shop scheduling problem optimally by dynamic programming
- A general VNS heuristic for the traveling salesman problem with time windows
- A restricted dynamic programming heuristic algorithm for the time dependent traveling salesman problem
- Tour merging via branch-decomposition
- A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play
- Three-dimensional matching instances are rich in stable matchings
- Generalization of machine learning for problem reduction: a case study on travelling salesman problems
- Reinforcement learning for combinatorial optimization: a survey
- New benchmark instances for the capacitated vehicle routing problem
- Deep learning assisted heuristic tree search for the container pre-marshalling problem
- Dynamic programming and optimal control. Vol. 1.
- Hybrid genetic search for the CVRP: open-source implementation and SWAP* neighborhood
Cited In (14)
- Learn and route: learning implicit preferences for vehicle routing
- Learning heuristics for the TSP by policy gradient
- Solving the traveling salesperson problem with precedence constraints by deep reinforcement learning
- Neural networks for local search and crossover in vehicle routing: a possible overkill?
- Deep infeasibility exploration method for vehicle routing problems
- A reinforcement learning approach to convoy scheduling on a contested transportation network
- Learning to repeatedly solve routing problems
- Solving large-scale routing optimization problems with networks and only networks
- Dynamic graph conv-LSTM model with dynamic positional encoding for the large-scale traveling salesman problem
- Research on improved ant colony optimization for traveling salesman problem
- Predicting the optimal period for Cyclic Hoist Scheduling Problems
- Opportunities for reinforcement learning in stochastic dynamic vehicle routing
- Neural large neighborhood search for routing problems
- Learning the travelling salesperson problem requires rethinking generalization
Uses Software
This page was built for publication: Deep policy dynamic programming for vehicle routing problems
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2170197)
