Neural networks for local search and crossover in vehicle routing: A possible overkill?. Santana, Ί., Lodi, A., & Vidal, T. In To appear in CPAIOR 2023, 2023. arXiv:2210.12075.
Paper abstract bibtex Extensive research has been conducted, over recent years, on various ways of enhancing heuristic search for combinatorial optimization problems with machine learning algorithms. In this study, we investigate the use of predictions from graph neural networks (GNNs) in the form of heatmaps to improve the Hybrid Genetic Search (HGS), a state-of-the-art algorithm for the Capacitated Vehicle Routing Problem (CVRP). The crossover and local-search components of HGS are instrumental in finding improved solutions, yet these components essentially rely on simple greedy or random choices. It seems intuitive to attempt to incorporate additional knowledge at these levels. Throughout a vast experimental campaign on more than 10,000 problem instances, we show that exploiting more sophisticated strategies using measures of node relatedness (heatmaps, or simply distance) within these algorithmic components can significantly enhance performance. However, contrary to initial expectations, we also observed that heatmaps did not present significant advantages over simpler distance measures for these purposes. Therefore, we faced a common – though rarely documented – situation of overkill: GNNs can indeed improve performance on an important optimization task, but an ablation analysis demonstrated that simpler alternatives perform equally well.
@inproceedings{Santana2023,
abstract = {Extensive research has been conducted, over recent years, on various ways of enhancing heuristic search for combinatorial optimization problems with machine learning algorithms. In this study, we investigate the use of predictions from graph neural networks (GNNs) in the form of heatmaps to improve the Hybrid Genetic Search (HGS), a state-of-the-art algorithm for the Capacitated Vehicle Routing Problem (CVRP). The crossover and local-search components of HGS are instrumental in finding improved solutions, yet these components essentially rely on simple greedy or random choices. It seems intuitive to attempt to incorporate additional knowledge at these levels. Throughout a vast experimental campaign on more than 10,000 problem instances, we show that exploiting more sophisticated strategies using measures of node relatedness (heatmaps, or simply distance) within these algorithmic components can significantly enhance performance. However, contrary to initial expectations, we also observed that heatmaps did not present significant advantages over simpler distance measures for these purposes. Therefore, we faced a common -- though rarely documented -- situation of overkill: GNNs can indeed improve performance on an important optimization task, but an ablation analysis demonstrated that simpler alternatives perform equally well.},
archivePrefix = {arXiv},
arxivId = {2210.12075},
author = {Santana, {\'{I}}. and Lodi, A. and Vidal, T.},
booktitle = {To appear in CPAIOR 2023},
eprint = {2210.12075},
file = {:C$\backslash$:/Users/Thibaut/Documents/Mendeley-Articles/Santana, Lodi, Vidal/Santana, Lodi, Vidal - 2023 - Neural networks for local search and crossover in vehicle routing A possible overkill.pdf:pdf},
organization = {arXiv:2210.12075},
title = {{Neural networks for local search and crossover in vehicle routing: A possible overkill?}},
url = {https://arxiv.org/pdf/2210.12075.pdf},
year = {2023}
}
Downloads: 0
{"_id":"NzrZC3e3bMAyg6oiX","bibbaseid":"santana-lodi-vidal-neuralnetworksforlocalsearchandcrossoverinvehicleroutingapossibleoverkill-2023","author_short":["Santana, Ί.","Lodi, A.","Vidal, T."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","abstract":"Extensive research has been conducted, over recent years, on various ways of enhancing heuristic search for combinatorial optimization problems with machine learning algorithms. In this study, we investigate the use of predictions from graph neural networks (GNNs) in the form of heatmaps to improve the Hybrid Genetic Search (HGS), a state-of-the-art algorithm for the Capacitated Vehicle Routing Problem (CVRP). The crossover and local-search components of HGS are instrumental in finding improved solutions, yet these components essentially rely on simple greedy or random choices. It seems intuitive to attempt to incorporate additional knowledge at these levels. Throughout a vast experimental campaign on more than 10,000 problem instances, we show that exploiting more sophisticated strategies using measures of node relatedness (heatmaps, or simply distance) within these algorithmic components can significantly enhance performance. However, contrary to initial expectations, we also observed that heatmaps did not present significant advantages over simpler distance measures for these purposes. Therefore, we faced a common – though rarely documented – situation of overkill: GNNs can indeed improve performance on an important optimization task, but an ablation analysis demonstrated that simpler alternatives perform equally well.","archiveprefix":"arXiv","arxivid":"2210.12075","author":[{"propositions":[],"lastnames":["Santana"],"firstnames":["Ί."],"suffixes":[]},{"propositions":[],"lastnames":["Lodi"],"firstnames":["A."],"suffixes":[]},{"propositions":[],"lastnames":["Vidal"],"firstnames":["T."],"suffixes":[]}],"booktitle":"To appear in CPAIOR 2023","eprint":"2210.12075","file":":C$\\$:/Users/Thibaut/Documents/Mendeley-Articles/Santana, Lodi, Vidal/Santana, Lodi, Vidal - 2023 - Neural networks for local search and crossover in vehicle routing A possible overkill.pdf:pdf","organization":"arXiv:2210.12075","title":"Neural networks for local search and crossover in vehicle routing: A possible overkill?","url":"https://arxiv.org/pdf/2210.12075.pdf","year":"2023","bibtex":"@inproceedings{Santana2023,\nabstract = {Extensive research has been conducted, over recent years, on various ways of enhancing heuristic search for combinatorial optimization problems with machine learning algorithms. In this study, we investigate the use of predictions from graph neural networks (GNNs) in the form of heatmaps to improve the Hybrid Genetic Search (HGS), a state-of-the-art algorithm for the Capacitated Vehicle Routing Problem (CVRP). The crossover and local-search components of HGS are instrumental in finding improved solutions, yet these components essentially rely on simple greedy or random choices. It seems intuitive to attempt to incorporate additional knowledge at these levels. Throughout a vast experimental campaign on more than 10,000 problem instances, we show that exploiting more sophisticated strategies using measures of node relatedness (heatmaps, or simply distance) within these algorithmic components can significantly enhance performance. However, contrary to initial expectations, we also observed that heatmaps did not present significant advantages over simpler distance measures for these purposes. Therefore, we faced a common -- though rarely documented -- situation of overkill: GNNs can indeed improve performance on an important optimization task, but an ablation analysis demonstrated that simpler alternatives perform equally well.},\narchivePrefix = {arXiv},\narxivId = {2210.12075},\nauthor = {Santana, {\\'{I}}. and Lodi, A. and Vidal, T.},\nbooktitle = {To appear in CPAIOR 2023},\neprint = {2210.12075},\nfile = {:C$\\backslash$:/Users/Thibaut/Documents/Mendeley-Articles/Santana, Lodi, Vidal/Santana, Lodi, Vidal - 2023 - Neural networks for local search and crossover in vehicle routing A possible overkill.pdf:pdf},\norganization = {arXiv:2210.12075},\ntitle = {{Neural networks for local search and crossover in vehicle routing: A possible overkill?}},\nurl = {https://arxiv.org/pdf/2210.12075.pdf},\nyear = {2023}\n}\n","author_short":["Santana, Ί.","Lodi, A.","Vidal, T."],"key":"Santana2023","id":"Santana2023","bibbaseid":"santana-lodi-vidal-neuralnetworksforlocalsearchandcrossoverinvehicleroutingapossibleoverkill-2023","role":"author","urls":{"Paper":"https://arxiv.org/pdf/2210.12075.pdf"},"metadata":{"authorlinks":{}},"html":""},"bibtype":"inproceedings","biburl":"https://w1.cirrelt.ca/~vidalt/resources/My Collection.bib","dataSources":["yinfondEAJRbDM9sJ","sempRA6PhmAdGk3yG"],"keywords":[],"search_terms":["neural","networks","local","search","crossover","vehicle","routing","possible","overkill","santana","lodi","vidal"],"title":"Neural networks for local search and crossover in vehicle routing: A possible overkill?","year":2023}