Pith. sign in

REVIEW 5 major objections 5 minor 71 references

USPR: Learning a Unified Solver for Profiled Routing

T0 review · 5 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A single reinforcement-learning-trained transformer, USPR, handles arbitrary vehicle-client preference profiles and preference weights without retraining while reporting better solution quality than prior learning-based PVRP solvers on…

desk verdict A credible single-model PVRP solver with solid in-distribution and scale gains, but the 'arbitrary profile types' headline claim is untested — every experiment uses one Uniform(0,1) profile generator. read the letter →

arxiv 2505.05119 v2 pith:JAADJW4U submitted 2025-05-08 cs.LG cs.MA

Chuanbo Hua , Federico Berto , Zhikai Zhao , Jiwoo Son , Changhyun Kwon , Jinkyoo Park This is my paper · ORCID
classification cs.LGcs.MA
keywords profiledvehicleroutingproblemunifiedneuralsolvermulti-headattentionprofileembeddingsscorereshapingreinforcementlearningout-of-distributiongeneralizationvehicle-clientpreferences
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

USPR is a single transformer-based policy trained once to solve the Profiled Vehicle Routing Problem (PVRP), where each vehicle-client pair has a preference score that can also encode hard constraints. The paper claims that one USPR model replaces the previous practice of training ten separate models on the same profile distribution, because its profile embeddings make the policy continuous in the profile matrix and in the preference weight $\alpha$. Across in-distribution instances with 60-100 clients and out-of-distribution instances up to 1,000 clients, USPR reports a lower average gap to the HGS-PyVRP reference than all compared learning-based solvers, at a fraction of the training cost. The paper's contribution is a unified neural construction policy for preference-aware routing, not a new classical algorithm.

What carries the argument

Three components carry the argument. Profile Embeddings (PE) project clients, vehicles, profile scores (including $\pm\infty$ hard constraints) and the scalar weight $\alpha$ into a shared latent space so the model can condition on any profile matrix. Multi-Head Profiled Attention (MHPA) runs four attention directions (client-client, vehicle-vehicle, vehicle-client, client-vehicle) to build bidirectional vehicle-client representations. Profile-aware Score Reshaping (PSR) adds an explicit distance-plus-preference term to the decoder logits, so the policy's outputs stay calibrated when profile distributions or instance sizes shift. Together they let a single autoregressive, parallel-vehicle decoder produce feasible routes without per-distribution retraining.

What would settle it

Generate a test set of PVRP instances whose profile scores come from a different distribution family, such as correlated preferences, ordinal categories, or real driver-customer ratings, and measure USPR's greedy gap to HGS-PyVRP; if the gap degrades sharply or falls behind CAMP on the same instances, the 'arbitrary profile types' unification claim would be refuted.

Watch

Extended reading notes

Core claim

The paper's central claim is that a single model, USPR, can natively handle arbitrary vehicle-client profile matrices and profile weights without retraining, and that it does so while being the best learning-based PVRP solver in the comparisons. The authors report that USPR's greedy solutions have a 9.20% average gap to the HGS-PyVRP reference on in-distribution instances of 60-100 clients, against 10.50% for the strongest baseline CAMP, and a 19.35% gap at 1,000 clients against 23.72% for CAMP. In the same comparison, one USPR model replaces ten CAMP models trained at fixed profile weights, cutting parameter count, epochs, and wall-clock training time by roughly an order of magnitude. The paper therefore positions USPR as a unification result: profile-aware routing can be handled by a single construction policy rather than a zoo of task-specific ones.

Load-bearing premise

Every benchmark, including the real-world PVRPLib set, generates profile scores from the same synthetic protocol — independent Uniform(0,1) draws with zero to ten percent of entries randomly set to plus or minus infinity — and the baselines are adapted through an unspecified shared recipe; if genuinely different preference distributions or a different adaptation protocol change the ranking, the unified-solver claim is weaker than stated.

Editorial extensions

If this is right

  • Deploying USPR removes the need to train and store a separate model for each preference weight or zone-constraint configuration.
  • On the paper's benchmarks, USPR's greedy solutions are closer to the HGS-PyVRP reference than those of the best prior neural baseline CAMP, with reported gaps of 9.20% versus 10.50% at N=60-100.
  • The reported gap advantage grows at out-of-distribution sizes, reaching 19.35% versus 23.72% at N=1000 in greedy mode.
  • One USPR model (1.5 million parameters, 100 epochs, about 11 hours) is reported to replace ten CAMP models (17.6 million parameters, 1000 epochs, about 4.6 days).

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct test the paper does not run is to change the profile distribution family at test time, for example correlated, categorical, or empirical driver-customer preferences; the architecture's embeddings should be distribution-agnostic, but no evidence yet shows this.
  • Because the reshaping term exposes distance and preference scales directly in the logits, the mechanism may also transfer across coordinate scales or non-Euclidean distance metrics, which the paper does not evaluate.
  • The memory limitation the paper reports, that profile matrix embeddings explode past roughly 10,000 clients, suggests a sparse or factorized profile-encoding variant as an obvious next step.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper proposes USPR, a unified neural solver for the Profiled Vehicle Routing Problem (PVRP), in which a single model trained once handles arbitrary vehicle-client profile matrices and profile weights without retraining. The method combines Profile Embeddings (PE) to encode profiles and weights, Multi-Head Profiled Attention (MHPA) for bidirectional vehicle-client interactions, and Profile-aware Score Reshaping (PSR) that injects explicit distance and profile information into decoder logits. The authors report in-distribution results on N=60-100, out-of-distribution scale generalization up to N=1000, and results on a newly constructed PVRPLib dataset; they also report an ablation and an efficiency comparison showing one USPR model replaces ten CAMP models. The central empirical claim is state-of-the-art performance among learning-based PVRP solvers, with a single model generalizing to instances up to 10x larger than training data.

Significance. The paper addresses a real limitation of existing neural PVRP solvers, which typically require separate models for different profile distributions and weights. The architectural ideas (PE, MHPA, PSR) are sensible and the experimental results consistently show USPR outperforming the neural baselines, including CAMP, across all tested scales, while being substantially cheaper to train. The scale-generalization numbers (e.g., Table 3, N=1000) are notable and support the method's practical potential. If the unified-solver claim could be substantiated across genuinely varied profile distributions, this would be a meaningful contribution to the NCO literature. However, the current evidence is narrower than the claim: all experiments use one synthetic profile generation protocol, and the absence of variance estimates and code availability limits the strength of the empirical conclusions.

major comments (5)
  1. [Experiments: Data Generation / Supplementary Data Generation] The central claim of a unified solver for 'arbitrary profile types' is not tested. In every experiment, including PVRPLib, profile scores are drawn independently from Uniform(0,1) with between 0% and 10% of entries set to ±infinity. No experiment varies the profile distribution family (e.g., categorical, correlated, negative-valued, or real driver-customer preferences), and the zone-constrained variant described in the supplementary is not evaluated in any reported table. Since the claim of handling arbitrary profile types is the paper's key selling point, the evidence currently supports only the narrow case of i.i.d. uniform soft scores with sparse hard constraints. The authors should either add experiments with substantially different profile distributions or explicitly restrict the scope claim in the abstract and introduction.
  2. [Tables 1, 3, 4, 5, 8, 9] No table reports variance or multiple-seed results. All neural methods are evaluated on a single run, so the observed gaps between USPR and the strongest baseline CAMP (e.g., Table 1 greedy gap 9.20% vs. 10.50%; Table 3 greedy gap 19.35% vs. 23.72%) could be within run-to-run noise. The paper's central empirical claim is that USPR achieves state-of-the-art results among learning-based methods; this claim requires at least a few independent training runs with reported means and standard deviations, or statistical significance tests.
  3. [Abstract and Supplementary 'Code Implementation and Hardware'] The abstract states that source code is publicly available, but the manuscript contains no link or repository identifier. The supplementary says only that code 'will be released upon acceptance.' This breaks the reproducibility expectation for a machine learning paper and makes it impossible to verify the reported results or baseline adaptations. The authors should provide an anonymous or working link to code and trained models, or clearly state in the abstract that code will be released upon publication.
  4. [Experiments: Neural Baselines] The statement 'We follow CAMP to adapt ET, DPN, 2D-Ptr, and PARCO to PVRP' is not substantiated. The paper gives no details on how profiles are incorporated into these architectures, whether hyperparameters were retuned, or whether the adaptation is equivalent across baselines. Without this information, the fairness of the comparison cannot be assessed. The authors should specify the adaptation protocol, ideally in an appendix.
  5. [Eq. (5), Profile-Aware Score Reshaping] The PSR formula uses log(dist_ij + p_ij). Since the paper's problem definition allows p_ik in R∪{±infinity}, this expression is undefined for p_ij = -infinity and for any p_ij ≤ -dist_ij, and it is also potentially ill-defined for +infinity. The paper does not explain how hard constraints are handled in the logit computation before masking, nor how the formula generalizes to negative or arbitrary real-valued profiles. This is a technical gap directly related to the 'arbitrary profile types' claim.
minor comments (5)
  1. [Introduction / Methodology] There is an inconsistency in component naming: the Introduction contribution list says 'PAR to ensure robust generalization', while the Methodology section and all experiments refer to PSR (Profile-aware Score Reshaping). Please correct the typo.
  2. [Table 3 caption] The caption claims results for instances 'up to 10× the number of agents M and 10× the number of nodes N'. While N=1000 is 10x the training maximum of N=100, the agent number M=35 is only 5x the training maximum of M=7. Please correct the factor for M.
  3. [Table 5] The ablation labels '-PSR & SR' and '-PSR, SR & MHPA' are unclear; 'SR' is not defined in the main text. Please state clearly what SR (distance-only score reshaping?) denotes and what each ablation removes.
  4. [Conclusion] The conclusion states that USPR is 'matching or exceeding classical solvers', but the reported gaps to HGS-PyVRP are consistently positive (e.g., 9.20% in Table 1 and 19.35% in Table 3). The paper only outperforms OR-Tools in some settings, not HGS-PyVRP. Please rephrase to avoid overclaiming.
  5. [Table 4] PVRPLib is described as 'real-world' data, but only the coordinates, demands, and vehicle counts come from CVRPLib; the profile scores are generated with the same synthetic Uniform(0,1) protocol. The terminology 'real-world' is misleading and should be clarified, e.g., 'real-world geometry with synthetic profiles'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: USPR's reported gains are benchmarked against external solvers, and its components are hand-designed rather than fit to the test set.

full rationale

USPR's empirical claims are checked against external solvers (HGS-PyVRP, OR-Tools) and prior neural methods, so the reported gaps are not produced by USPR's own assumptions. The three components PE, MHPA, and PSR are introduced as architectural design choices (Eqs. 2-5), not as parameters fitted to the reported test instances; PSR's injection of distance and profile scores into logits is a hand-specified heuristic (the paper explicitly notes 'we manually design the attention reshaping mechanism'), and the policy is trained with REINFORCE on the same objective, so evaluation does not reduce to a training artifact. The main neural baselines CAMP and PARCO are from the same research group, but they are used as external comparisons and are not cited to establish USPR's correctness. The sentence 'We follow CAMP to adapt ET, DPN, 2D-Ptr, and PARCO to PVRP for a fair comparison' is underspecified, but it concerns baseline implementation rather than the derivation of USPR's central results. The 'arbitrary profile types' claim is broader than the evidence: every experiment samples profiles i.i.d. from Uniform(0,1) with 0-10% of entries set to ±infinity (main text and supplementary), and PVRPLib reuses that same profile generator; however, an untested generalization claim is an evidential gap, not a circular derivation. No equation or claimed prediction reduces by construction to an input or to a same-author result.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

USPR introduces no speculative physical entities. The ledger items that matter are design choices: the hand-built PSR formula, the alpha range [0, 0.2], the hard-constraint sampling rule, the baseline adaptation inherited from CAMP, and the single synthetic profile distribution used in all benchmarks. None of these were fitted against the test sets, so the direct comparisons remain informative, but they bound the breadth of the 'arbitrary profile types' claim.

free parameters (4)
  • PSR log-space combination log(dist_ij + p_ij)
    Eq. (5) adds distance and profile score inside one logarithm with equal weighting; the paper gives no tuning or derivation for this form, so it is a hand-chosen inductive bias that the ablation shows improves results.
  • Logits scaling constant C = 10
    Set to 10 following Bello et al. (2016); this temperature-like factor shapes the policy entropy and is not re-tuned for PVRP.
  • Training range of profile weight alpha = [0, 0.2]
    alpha is sampled uniformly from [0, 0.2] during training and all quantitative benchmarks average over this range only; Fig. 3's alpha=1 demonstration is outside the trained range.
  • Hard-constraint sampling probabilities = two draws from Uniform(0, 0.1)
    Each instance randomly converts zero to ten percent of client-vehicle pairs to -infinity or +infinity; this ad hoc choice defines every benchmark's profile distribution.
assumptions (4)
  • domain assumption The MDP reward (Eq. 11) correctly implements the ILP objective (Eq. 1).
    Eq. (1) attaches the profile score to the departure node i while Eq. (11) attaches it to the arrival node j; the totals coincide for complete tours only if the depot profile p_0k is defined, which the paper never does. This is a definitional gap, not a measured one.
  • domain assumption Adapting ET, DPN, 2D-Ptr and PARCO to PVRP via CAMP's protocol yields fair, strong baselines.
    The paper gives one sentence ('We follow CAMP to adapt...') with no implementation details; the SOTA claim depends on these baselines being representative of their best possible PVRP performance.
  • domain assumption Uniform(0,1) profile scores plus zero to ten percent random plus or minus infinity masks represent real-world PVRP profile distributions, including in PVRPLib.
    This single protocol generates every training and test instance, and PVRPLib overlays it on CVRPLib coordinates, so 'arbitrary profile types' and 'real-world' are not empirically stressed.
  • ad hoc to paper Including the explicit log(dist_ij + p_ij) penalty in decoder logits (PSR) improves generalization without distorting the problem objective.
    PSR is a manually designed heuristic term; its benefit is evidenced only by the ablation (4.42% to 4.72% gap), not by a derivation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of USPR: Learning a Unified Solver for Profiled Routing." pith.science (2026). https://pith.science/paper/JAADJW4U

@misc{pith2026250505119,
  author       = {Pith},
  title        = {Pith review of: USPR: Learning a Unified Solver for Profiled Routing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JAADJW4U}},
  note         = {Machine review of arXiv:2505.05119}
}
read the original abstract

The Profiled Vehicle Routing Problem (PVRP) extends the classical VRP by incorporating vehicle-client-specific preferences and constraints, reflecting real-world requirements such as zone restrictions and service-level preferences. While recent reinforcement-learning solvers have shown promising performance, they require retraining for each new profile distribution, suffer from poor representation ability, and struggle to generalize to out-of-distribution instances. In this paper, we address these limitations by introducing Unified Solver for Profiled Routing (USPR), a novel framework that natively handles arbitrary profile types. USPR introduces on three key innovations: (i) Profile Embeddings (PE) to encode any combination of profile types; (ii) Multi-Head Profiled Attention (MHPA), an attention mechanism that models rich interactions between vehicles and clients; (iii) Profile-aware Score Reshaping (PSR), which dynamically adjusts decoder logits using profile scores to improve generalization. Empirical results on diverse PVRP benchmarks demonstrate that USPR achieves state-of-the-art results among learning-based methods while offering significant gains in flexibility and computational efficiency. We make our source code publicly available to foster future research.

Figures

Figures reproduced from arXiv: 2505.05119 by the authors.

Figure 1
Figure 1. An illustrative example of the PVRP. The profiles [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. An overview of USPR. Our framework follows an encoder-decoder architecture and introduces three key components [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. USPR’s PVRP solutions to the same instance with [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

71 extracted references · 56 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Aiko, S.; Thaithatukl, P.; and Asakura, Y. 2018. Incorporating user preference into optimal vehicle routing problem of integrated sharing transport system. Asian Transport Studies, 5(1): 98--116

    work page 2018

  4. [4]

    K.; and Schmidt-Thieme, L

    Bdeir, A.; Falkner, J. K.; and Schmidt-Thieme, L. 2022. Attention, filling in the gaps for generalization in routing problems. In ECML PKDD, 505--520. Springer

    work page 2022

  5. [5]

    V.; Norouzi, M.; and Bengio, S

    Bello, I.; Pham, H.; Le, Q. V.; Norouzi, M.; and Bengio, S. 2016. Neural combinatorial optimization with reinforcement learning. arXiv preprint arXiv:1611.09940

    arXiv 2016

  6. [6]

    Bengio, Y.; Lodi, A.; and Prouvost, A. 2021. Machine learning for combinatorial optimization: a methodological tour d’horizon. European Journal of Operational Research, 290(2): 405--421

    2021

  7. [7]

    Berto, F.; Hua, C.; Luttmann, L.; Son, J.; Park, J.; Ahn, K.; Kwon, C.; Xie, L.; and Park, J. 2024 a . PARCO: Learning Parallel Autoregressive Policies for Efficient Multi-Agent Combinatorial Optimization . arXiv preprint arXiv:2409.03811

    arXiv 2024

  8. [8]

    Berto, F.; Hua, C.; Park, J.; Luttmann, L.; Ma, Y.; Bu, F.; Wang, J.; Ye, H.; Kim, M.; Choi, S.; Zepeda, N. G.; Hottung, A.; Zhou, J.; Bi, J.; Hu, Y.; Liu, F.; Kim, H.; Son, J.; Kim, H.; Angioni, D.; Kool, W.; Cao, Z.; Zhang, J.; Shin, K.; Wu, C.; Ahn, S.; Song, G.; Kwon, C.; Xie, L.; and Park, J. 2025. RL4CO: an Extensive Reinforcement Learning for Combi...

    work page 2025

Show all 71 references
  1. [9]

    G.; Hottung, A.; Wouda, N.; Lan, L.; Tierney, K.; and Park, J

    Berto, F.; Hua, C.; Zepeda, N. G.; Hottung, A.; Wouda, N.; Lan, L.; Tierney, K.; and Park, J. 2024 b . RouteFinder : Towards Foundation Models for Vehicle Routing Problems. In ICML 2024 FM-Wild Workshop

    2024

  2. [10]

    Bi, J.; Ma, Y.; Zhou, J.; Song, W.; Cao, Z.; Wu, Y.; and Zhang, J. 2024. Learning to Handle Complex Constraints for Vehicle Routing Problems. arXiv preprint arXiv:2410.21066

    2024 arXiv

  3. [11]

    Bogyrbayeva, A.; Yoon, T.; Ko, H.; Lim, S.; Yun, H.; and Kwon, C. 2023. A deep reinforcement learning approach for solving the traveling salesman problem with drone. Transportation Research Part C: Emerging Technologies, 148: 103981

    2023

  4. [12]

    Braekers, K.; Ramaekers, K.; and Van Nieuwenhuyse, I. 2016. The vehicle routing problem: State of the art classification and review. Computers & industrial engineering, 99: 300--313

    2016

  5. [13]

    Choo, J.; Kwon, Y.-D.; Kim, J.; Jae, J.; Hottung, A.; Tierney, K.; and Gwon, Y. 2022. Simulation-guided beam search for neural combinatorial optimization. NeurIPS, 35: 8760--8772

    2022

  6. [14]

    Cordeau, J.-F.; and Laporte, G. 2001. A tabu search heuristic for the site dependent vehicle routing problem with time windows. INFOR, 39(4): 292--298

    2001

  7. [15]

    Drakulic, D.; Michel, S.; and Andreoli, J.-M. 2025. GOAL: A Generalist Combinatorial Optimization Agent Learning. In ICLR

    2025

  8. [16]

    Drakulic, D.; Michel, S.; Mai, F.; Sors, A.; and Andreoli, J.-M. 2024. Bq-nco: Bisimulation quotienting for efficient neural combinatorial optimization. NeurIPS, 36

    2024

  9. [17]

    Duan, L.; Zhan, Y.; Hu, H.; Gong, Y.; Wei, J.; Zhang, X.; and Xu, Y. 2020. Efficiently Solving the Practical Vehicle Routing Problem : A Novel Joint Learning Approach . In KDD. ACM

    2020

  10. [18]

    L.; Ma, Y.; Zhou, J.; Cao, Z.; Dupty, M

    Goh, Y. L.; Ma, Y.; Zhou, J.; Cao, Z.; Dupty, M. H.; and Lee, W. S. 2025. SHIELD: Multi-task Multi-distribution Vehicle Routing Solver with Sparsity & Hierarchy in Efficiently Layered Decoder. In ICML

    2025

  11. [19]

    Golden, B.; Assad, A.; Levy, L.; and Gheysens, F. 1984. The fleet size and mix vehicle routing problem. Computers & Operations Research, 11(1): 49--66

    1984

  12. [20]

    Grinsztajn, N.; Furelos-Blanco, D.; Surana, S.; Bonnet, C.; and Barrett, T. 2023. Winner takes it all: Training performant RL populations for combinatorial optimization. NeurIPS, 36: 48485--48509

    2023

  13. [21]

    Grinsztajn, N.; Furelos-Blanco, D.; Surana, S.; Bonnet, C.; and Barrett, T. 2024. Winner Takes It All: Training Performant RL Populations for Combinatorial Optimization. NeurIPS, 36

    2024

  14. [22]

    Hottung, A.; Kwon, Y.-D.; and Tierney, K. 2022. Efficient active search for combinatorial optimization problems. In ICLR

    2022

  15. [23]

    Hottung, A.; Mahajan, M.; and Tierney, K. 2025. PolyNet : Learning Diverse Solution Strategies for Neural Combinatorial Optimization. ICLR

    2025

  16. [24]

    Hottung, A.; and Tierney, K. 2020. Neural large neighborhood search for the capacitated vehicle routing problem. In ECAI 2020. IOS Press

    2020

  17. [25]

    Hottung, A.; Wong-Chung, P.; and Tierney, K. 2025. Neural Deconstruction Search for Vehicle Routing Problems. TMLR

    2025

  18. [26]

    Hua, C.; Berto, F.; Son, J.; Kang, S.; Kwon, C.; and Park, J. 2025. CAMP: Collaborative Attention Model with Profiles for Vehicle Routing Problems . In AAMAS

    2025

  19. [27]

    Huang, Z.; Zhou, J.; Cao, Z.; and Xu, Y. 2025. Rethinking Light Decoder-based Solvers for Vehicle Routing Problems. arXiv preprint arXiv:2503.00753

    2025 arXiv

  20. [28]

    S.; and McGeoch, L

    Johnson, D. S.; and McGeoch, L. A. 1997. The traveling salesman problem: a case study. Local search in combinatorial optimization

    1997

  21. [29]

    Kim, H.; Choi, S.; Son, J.; Park, J.; and Kwon, C. 2025. Neural Genetic Search in Discrete Spaces. In ICML

    2025

  22. [30]

    Kim, M.; Choi, S.; Son, J.; Kim, H.; Park, J.; and Bengio, Y. 2024. Ant Colony Sampling with GFlowNets for Combinatorial Optimization. arXiv preprint arXiv:2403.07041

    2024 arXiv

  23. [31]

    Kim, M.; Park, J.; and Park, J. 2022 a . Neuro CROSS exchange: Learning to CROSS exchange to solve realistic vehicle routing problems. arXiv preprint arXiv:2206.02771

    2022 arXiv

  24. [32]

    Kim, M.; Park, J.; and Park, J. 2022 b . Sym-nco: Leveraging symmetricity for neural combinatorial optimization. NeurIPS, 35: 1936--1949

    2022

  25. [33]

    P.; and Ba, J

    Kingma, D. P.; and Ba, J. 2014. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980

    2014 arXiv

  26. [34]

    Kool, W.; van Hoof, H.; Gromicho, J.; and Welling, M. 2022. Deep policy dynamic programming for vehicle routing problems. In CPAIOR, 190--213. Springer

    2022

  27. [35]

    Kool, W.; Van Hoof, H.; and Welling, M. 2018. Attention, learn to solve routing problems! arXiv preprint arXiv:1803.08475

    2018 arXiv

  28. [36]

    Kwon, Y.-D.; Choo, J.; Kim, B.; Yoon, I.; Gwon, Y.; and Min, S. 2020. Pomo: Policy optimization with multiple optima for reinforcement learning. NeurIPS, 33: 21188--21198

    2020

  29. [37]

    Li, H.; Liu, F.; Zheng, Z.; Zhang, Y.; and Wang, Z. 2024 a . CaDA: Cross-Problem Routing Solver with Constraint-Aware Dual-Attention. arXiv preprint arXiv:2412.00346

    2024 arXiv

  30. [38]

    Li, J.; Ma, Y.; Gao, R.; Cao, Z.; Lim, A.; Song, W.; and Zhang, J. 2022. Deep reinforcement learning for solving the heterogeneous capacitated vehicle routing problem. IEEE Transactions on Cybernetics

    2022

  31. [39]

    Li, S.; Yan, Z.; and Wu, C. 2021. Learning to delegate for large-scale vehicle routing. NeurIPS, 34: 26198--26211

    2021

  32. [40]

    Li, Y.; Guo, J.; Wang, R.; and Yan, J. 2024 b . From distribution learning in training to gradient search in testing for combinatorial optimization. NeurIPS, 36

    2024

  33. [41]

    Li, Y.; Ma, J.; Pan, W.; Wang, R.; Geng, H.; Yang, N.; and Yan, J. 2025. Unify ml4tsp: Drawing methodological principles for tsp and beyond from streamlined design space of learning and search. In The Thirteenth International Conference on Learning Representations

    2025

  34. [42]

    Li, Y.; Zhou, C.; Yuan, P.; and Ngo, T. T. A. 2023. Experience-based territory planning and driver assignment with predicted demand and driver present condition. Transportation research part E: logistics and transportation review, 171: 103036

    2023

  35. [43]

    Liu, F.; Lin, X.; Wang, Z.; Zhang, Q.; Xialiang, T.; and Yuan, M. 2024 a . Multi-task learning for routing problem with cross-problem zero-shot generalization. In KDD

    2024

  36. [44]

    Liu, F.; Xialiang, T.; Yuan, M.; Lin, X.; Luo, F.; Wang, Z.; Lu, Z.; and Zhang, Q. 2024 b . Evolution of Heuristics: Towards Efficient Automatic Algorithm Design Using Large Language Model. In ICML

    2024

  37. [45]

    Liu, Q.; Liu, C.; Niu, S.; Long, C.; Zhang, J.; and Xu, M. 2024 c . 2D-Ptr: 2D Array Pointer Network for Solving the Heterogeneous Capacitated Vehicle Routing Problem. In AAMAS, 1238--1246

    2024

  38. [46]

    Lozano, M.; Molina, D.; and Herrera, F. 2011. Editorial scalability of evolutionary algorithms and other metaheuristics for large-scale continuous optimization problems. Soft computing, 15: 2085--2087

    2011

  39. [47]

    Ma, Y.; Cao, Z.; and Chee, Y. M. 2024. Learning to search feasible and infeasible regions of routing problems with flexible neural k-opt. NeurIPS, 36

    2024

  40. [48]

    Mazyavkina, N.; Sviridov, S.; Ivanov, S.; and Burnaev, E. 2021. Reinforcement learning for combinatorial optimization: A survey. Computers & Operations Research, 134: 105400

    2021

  41. [49]

    Ouyang, W.; Li, S.; Ma, Y.; and Wu, C. 2025. Learning to Segment for Capacitated Vehicle Routing Problems

    2025

  42. [50]

    H.; and Steiglitz, K

    Papadimitriou, C. H.; and Steiglitz, K. 1998. Combinatorial optimization: algorithms and complexity. Courier Corporation

    1998

  43. [51]

    Perron, L.; and Furnon, V. 2023. OR-Tools . Google

    2023

  44. [52]

    Pham, V. T. D.; Doan, L.; and Huynh, T. T. B. 2025. HSEvo: Elevating Automatic Heuristic Design with Diversity-Driven Harmony Search and Genetic Algorithm Using LLMs . In AAAI. Association for the Advancement of Artificial Intelligence (AAAI)

    2025

  45. [53]

    Pirnay, J.; and Grimm, D. G. 2024. Take a step and reconsider: Sequence decoding for self-improved neural combinatorial optimization. In ECAI

    2024

  46. [54]

    Son, J.; Kim, M.; Choi, S.; Kim, H.; and Park, J. 2024. Equity-Transformer: Solving NP-Hard Min-Max Routing Problems as Sequential Generation with Equity Context. In AAAI

    2024

  47. [55]

    Son, J.; Zhao, Z.; Berto, F.; Hua, C.; Kwon, C.; and Park, J. 2025. Neural Combinatorial Optimization for Real-World Routing. arXiv preprint arXiv:2503.16159

    2025

  48. [56]

    Sun, Z.; Li, Z.; Wang, H.; He, D.; Lin, Z.; and Deng, Z. 2019. Fast structured decoding for sequence models. NeurIPS, 32

    2019

  49. [57]

    Sun, Z.; and Yang, Y. 2024. Difusco: Graph-based diffusion solvers for combinatorial optimization. NeurIPS, 36

    2024

  50. [58]

    Team Locus . 2020. Zone-Based Routing is the Need of the Hour. Locus Blog. Access: 2024-10-17

    2020

  51. [59]

    D.; Nguyen-Tri, Q.; Binh, H

    Tran, C. D.; Nguyen-Tri, Q.; Binh, H. T. T.; and Thanh-Tung, H. 2025. Large Language Models powered Neural Solvers for Generalized Vehicle Routing Problems. In Towards Agentic AI for Science: Hypothesis Generation, Comprehension, Quantification, and Validation

    2025

  52. [60]

    Vidal, T. 2022. Hybrid genetic search for the CVRP: Open-source implementation and SWAP* neighborhood. Computers & Operations Research, 140: 105643

    2022

  53. [61]

    Vinyals, O.; Fortunato, M.; and Jaitly, N. 2015. Pointer networks. NeurIPS, 28

    2015

  54. [62]

    Wang, Y.; Jia, Y.-H.; Chen, W.-N.; and Mei, Y. 2024. Distance-aware Attention Reshaping: Enhance Generalization of Neural Solver for Large-scale Vehicle Routing Problems. arXiv preprint arXiv:2401.06979

    2024 arXiv

  55. [63]

    A.; Lan, L.; and Kool, W

    Wouda, N. A.; Lan, L.; and Kool, W. 2024. P y VRP : A high-performance VRP solver package. INFORMS Journal on Computing

    2024

  56. [64]

    Ye, H.; Wang, J.; Cao, Z.; Berto, F.; Hua, C.; Kim, H.; Park, J.; and Song, G. 2024. ReEvo: Large Language Models as Hyper-Heuristics with Reflective Evolution. In NeurIPS

    2024

  57. [65]

    Ye, H.; Wang, J.; Cao, Z.; Liang, H.; and Li, Y. 2023. DeepACO: Neural-enhanced Ant Systems for Combinatorial Optimization. In NeurIPS

    2023

  58. [66]

    Zhang, N.; Yang, J.; Cao, Z.; and Chi, X. 2025. Adversarial Generative Flow Network for Solving Vehicle Routing Problems. arXiv preprint arXiv:2503.01931

    2025 arXiv

  59. [67]

    Zhao, Z.; Hua, C.; Berto, F.; Lee, K.; Ma, Z.; Li, J.; and Park, J. 2025. TrajEvo : Designing Trajectory Prediction Heuristics via LLM-driven Evolution. arXiv preprint arXiv:2505.04480

    2025 arXiv

  60. [68]

    Zheng, Z.; Yao, S.; Wang, Z.; Xialiang, T.; Yuan, M.; and Tang, K. 2024. DPN : Decoupling Partition and Navigation for Neural Solvers of Min-max Vehicle Routing Problems. In ICML

    2024

  61. [69]

    W.; and Dessouky, M

    Zhong, H.; Hall, R. W.; and Dessouky, M. 2007. Territory planning and vehicle dispatching with driver learning. Transportation Science

    2007

  62. [70]

    Zhou, J.; Cao, Z.; Wu, Y.; Song, W.; Ma, Y.; Zhang, J.; and Xu, C. 2024. MVMoE: Multi-Task Vehicle Routing Solver with Mixture-of-Experts. In ICML

    2024

  63. [71]

    Zong, Z.; Zheng, M.; Li, Y.; and Jin, D. 2022. Mapdp: Cooperative multi-agent reinforcement learning to solve pickup and delivery problems. In AAAI

    2022

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.