Pith. sign in

REVIEW 3 major objections 6 minor 40 references

Automated architecture search can design traffic prediction models, and one-shot methods currently lead on cost and accuracy.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 15:14 UTC pith:VWIAJL2D

load-bearing objection Useful taxonomy, shaky comparative claims: the survey organizes NAS-for-traffic well, but its 'best balance' verdict for one-shot methods doesn't follow from its own table. the 3 major comments →

arxiv 2607.26467 v1 pith:VWIAJL2D submitted 2026-07-29 cs.LG cs.NE

Neural Architecture Search for Traffic Prediction: A Survey of Methods, Challenges, and Future Directions

Truong Giang Vu , Li Yang , Richard W. Pazzi This is my paper
classification cs.LG cs.NE
keywords neural architecture searchtraffic predictionspatio-temporal forecastinggraph neural networksone-shot NASgradient-based NASevolutionary NASautomated machine learning
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This survey reviews roughly ten neural architecture search (NAS) methods applied to traffic prediction, grouping them by search strategy: gradient-based, evolutionary, and one-shot weight-sharing. Its central claim is that one-shot methods — which train a shared supernet and sample sub-architectures from it — currently offer the best balance of search cost and prediction accuracy on standard benchmarks. A sympathetic reader should care because traffic prediction models are usually designed by hand and often fail to transfer across cities; NAS promises to automate that design. The paper also maps five open challenges: computational cost, manual search space design, cross-city transfer, dynamic graphs, and NAS for spatial-temporal foundation models.

Core claim

The paper establishes a taxonomy of NAS strategies for spatio-temporal traffic data and argues that auto-found architectures frequently match or outperform hand-designed baselines. It identifies one-shot weight-sharing methods (AutoCTS, AutoCTS+, AutoSTF) as the current best cost-accuracy trade-off, because they train a single supernet and evaluate candidates by sampling subgraphs without retraining, lowering search time dramatically. It also argues that gradient-based methods are fast but memory-intensive and prone to architecture collapse, while evolutionary methods avoid gradient bias but require many costly evaluations on graph-structured data.

What carries the argument

The taxonomy itself: (1) gradient-based differentiable search, (2) evolutionary population search, (3) one-shot weight-sharing supernet search. The search space design — which spatial operations (graph convolution, learnable adjacency) and temporal operations (LSTM, dilated convolution, attention) are candidates — is the central object that determines what architectures can be found. The one-shot supernet is the mechanism that lowers cost by sharing weights across candidates.

Load-bearing premise

The comparative conclusion that one-shot methods offer the best balance depends on the per-method MAE and GPU-hour numbers in Table II being mutually reliable, but the table itself warns that MAE values are not directly comparable across entries.

What would settle it

A controlled benchmark that trains one gradient-based, one evolutionary, and one one-shot method on the same traffic datasets (e.g., METR-LA and PeMS-BAY) with equal search budgets and an identical evaluation protocol would falsify the survey's ranking if the one-shot method does not post the best cost-accuracy trade-off.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Practitioners seeking an automated traffic model should consider one-shot supernet methods as a starting point, since they offer the best reported cost-accuracy balance.
  • Gradient-based NAS, while fast, remains memory-intensive on large graphs and prone to degenerate 'skip-heavy' architectures, so it needs stability remedies.
  • Zero-cost proxy scoring is a promising complement to supernet training, potentially cutting evaluation cost from hours to seconds.
  • Decoupled search (searching spatial and temporal subspaces separately) reduces memory and search time while preserving quality.
  • The open challenges imply that NAS for traffic is not yet deployment-ready at city scale (e.g., 8,000+ sensors), pointing to a need for scalability research.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The survey's table note states MAE values are not directly comparable across methods, yet the conclusion ranks one-shot methods best; a strict comparative ranking cannot be supported by the table's own numbers.
  • The survey's claim that gradient-based methods cost hundreds of GPU-hours conflicts with its own table listing 2-12 hours for those methods; readers should treat cost claims cautiously.
  • Cross-city generalization is the real bottleneck; NAS could be extended from searching a fixed architecture to searching a transferable initialization via meta-learning across cities.
  • A testable extension: run a controlled benchmark with equal training budgets across the three strategy families on the same datasets and metrics to verify the cost-accuracy ranking.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. This survey reviews neural architecture search (NAS) methods for traffic prediction, organizing roughly ten recent methods into gradient-based, evolutionary, and one-shot/weight-sharing families (with a brief discussion of RL and Bayesian approaches). It connects NAS search-space design to spatial-temporal traffic data challenges, provides a summary table with benchmark MAE and approximate GPU-hour search costs, and discusses five open challenges. The paper's central comparative claim, stated in Section V and Section III.C.3, is that one-shot methods such as AutoCTS and AutoSTF currently offer the best cost-accuracy balance on standard benchmarks.

Significance. The survey fills a genuine gap: while there are surveys on general NAS, graph-based traffic forecasting, and AutoML for traffic, none focuses specifically on NAS search strategies and search spaces for spatio-temporal traffic data. The taxonomy is reasonable and the discussion of spatial and temporal operation candidates, decoupled search, and the coupling problem is informative. The compiled table of methods and costs is a useful starting resource if verified. However, the paper's practical payoff—the comparative ranking of one-shot methods—is undermined by internal inconsistencies between the table's own caveats and the conclusions drawn from it. The descriptive and organizational content is likely useful to practitioners, but the comparative guidance requires correction rather than being merely stylistically weak.

major comments (3)
  1. [Section V and Table II; also Section III.C.3] The concluding assertion that 'one-shot methods ... currently offer the best balance of cost and accuracy on standard benchmarks' is not supported by the evidence in Table II. The table's own note states that 'MAE values are not directly comparable across all entries' and that the table should be read as 'a contextual summary rather than a strict ranking.' On the only benchmark with multiple strategies represented, METR-LA, the one-shot methods are neither the most accurate nor the cheapest: AutoSTF achieves MAE 3.35 at ~25 GPU-h and AutoCTS achieves 3.47 at 21 GPU-h, while gradient-based AutoSTG+ achieves 3.00 at ~12 GPU-h and evolutionary LENAS achieves 4.00 at 1–4 GPU-h. The one-shot family occupies the middle of the cost-accuracy trade-off on that benchmark. The concluding claim and the similar statement at the end of Section III.C.3 must be rewritten to respect the non-comparability
  2. [Section IV.1 and Table III] The claim that 'Even efficient gradient-based methods require hundreds of GPU-hours on standard benchmarks' is directly contradicted by the paper's own Table II, which lists AutoSTG at ~10 GPU-h, AutoSTG+ at ~12 GPU-h, and AutoSTL at ~2 GPU-h. The citations provided for the 'hundreds of GPU-hours' statement—[11] and [18]—are to AutoCTS (a one-shot method, 21 GPU-h) and Klosa et al. (an evolutionary method, ~1200 GPU-h), respectively; neither supports a claim about gradient-based methods. This statement is repeated in Table III and reinforced in the conclusion. It should be corrected or removed, and the computational-cost discussion should be grounded in the numbers actually reported in Table II.
  3. [Table II, footnote c] The footnote states 'AutoCTS+ does not evaluate on METR-LA'. As far as the cited paper (Wu et al., PACMMOD 2023) reports, AutoCTS+ evaluates on METR-LA as one of its primary benchmarks. This factual inaccuracy in a table footnote undermines the reliability of the table. All rows and footnotes should be re-verified against the primary sources; the table is load-bearing for the paper's comparative conclusions, so these checks are not optional presentation details.
minor comments (6)
  1. [Section III.B] There are missing spaces in the phrases 'Amacrosearch space' and 'Acell-basedspace'; these are simple typos but affect readability.
  2. [Table II caption] The caption uses '—' and 'N/R' inconsistently; define both symbols explicitly in the footnote area for consistency.
  3. [Section II.A] The problem setup uses T and H (historical steps and prediction horizon) without explicitly defining them; add a sentence defining T and H to make the formulation self-contained.
  4. [Section III.A, Eq. (1)] The bi-level optimization formulation is correct in idea, but the displayed equation is visually broken in the text extraction; ensure the rendered version clearly separates the inner and outer problems.
  5. [Section I] The introduction states that papers were identified using Google Scholar, IEEE Xplore, and ACM Digital Library but gives no detail on inclusion/exclusion criteria. Adding a brief description of the screening process would improve reproducibility of the survey.
  6. [Section III.C.3] The description of AutoCTS+ as 'joint search over both the architecture and hyperparameters' would benefit from specifying which hyperparameters are searched (e.g., number of layers, learning rate), as reported in the original paper.

Circularity Check

0 steps flagged

No circular derivation found: the survey’s taxonomy is self-contained and the two author self-citations are background only. The apparent contradiction between Table II and the conclusion is an internal-consistency/correctness issue, not circularity.

full rationale

This is a survey with no derivation chain that reduces a purported prediction or first-principles result back to its inputs. The taxonomy (gradient-based, evolutionary, one-shot) is organized from the reviewed papers’ own descriptions and is not defined in terms of the survey’s conclusions. Eq. (1) is a standard bi-level formulation of NAS, and citing [36] for it is ordinary background sourcing, not a load-bearing self-citation. Ref. [25] is used only to position the survey relative to adjacent work; even if it is the authors’ own review, the gap claim does not depend on a theorem or fitted parameter supplied by that citation. The strongest comparative claim — that one-shot methods “currently offer the best balance of cost and accuracy” — is an evidentiary assessment of Table II. Table II itself warns that “MAE values are not directly comparable across all entries” and that the table “should be read as a contextual summary rather than a strict ranking,” while Section V states a strict ranking and Section IV.1 says gradient methods require “hundreds of GPU-hours” despite Table II listing gradient entries at 2–12 GPU-hours. This is a data-comparability and argument-support problem, not a circularity problem: the conclusion is unsupported or inconsistent with the table, but it is not constructed from the table’s numbers by definition. No self-definitional relationship, fitted-input-as-prediction, imported uniqueness theorem, or ansatz-over-citation pattern is present. Because the paper’s central content is independent of the authors’ prior work, with only minor non-load-bearing self-citations, the appropriate circularity score is within the 0–2 range; I assign 2 to acknowledge the background self-citations while finding no significant circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 0 invented entities

A survey introduces no fitted parameters. The ledger instead records the background assumptions the review depends on: the NAS bi-level formulation (Eq. 1), the DARTS relaxation as the basis for the gradient family, the benchmarks as the currency of comparison, and the unverified premise that the reviewed papers' self-reported results support the survey's comparative verdicts. No new entities are invented. The strongest dependency, as with any survey, is the trust placed in the primary papers' reported MAE/search-cost values, which Table II itself flags as not directly comparable.

axioms (5)
  • standard math The bi-level optimization formulation of NAS (Eq. 1) is an adequate formalization of the traffic-NAS problem.
    Invoked in Section III.A and attributed to [36]; the survey relies on this framing without deriving or questioning it.
  • domain assumption NAS-found architectures 'often match or outperform hand-designed models' as reported by the surveyed papers.
    Stated in Sections I and V, supported only by the self-reported benchmarks of the methods being reviewed; no independent re-evaluation is performed.
  • standard math Search-space size grows as (Ms x Mt)^L for L stacked layers.
    Section III.B; standard combinatorics, correct as stated.
  • domain assumption Traffic prediction quality is adequately captured by MAE/RMSE/MAPE on METR-LA, PeMS-BAY, and PeMS04/08.
    Section II.A; the survey's comparative statements inherit this benchmark assumption from the cited papers.
  • standard math The DARTS continuous relaxation is a valid basis for the gradient-based NAS family.
    Section III.C.1; background from [13], assumed without critical assessment in the traffic setting.

pith-pipeline@v1.3.0-daily-deepseek · 11785 in / 18966 out tokens · 174087 ms · 2026-08-01T15:14:41.081380+00:00 · methodology

0 comments
read the original abstract

Traffic prediction is a core task in intelligent transportation systems, supporting applications such as adaptive signal control, route guidance, and ride-hailing dispatch. Deep learning models, including graph convolutional networks, recurrent networks, and Transformers, achieve strong results on standard benchmarks, but their architectures are designed by hand, requiring significant expert effort and producing models that often generalize poorly across cities and datasets. Neural Architecture Search (NAS) offers a systematic alternative to manual design. It automates the search over candidate architectures of deep learning models, finding designs that match the spatial-temporal structure of traffic data without manual trial and error. This survey reviews NAS methods applied to traffic prediction, organized by search strategy: gradient-based methods, evolutionary methods, and one-shot weight-sharing methods. For each category, we analyze how the search space is designed to cover spatial and temporal traffic operators, and how the search strategy balances cost against architecture quality. We also discuss open challenges, computational scalability to large road networks, manual search space design, cross-city generalization, dynamic graph structure, and the open question of NAS for spatial-temporal foundation models, and identify directions for future research.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

40 extracted references · 1 linked inside Pith

  1. [1]

    Prediction based traffic management in a metropolitan area,

    S. Chavhan and P. Venkataram, “Prediction based traffic management in a metropolitan area,”Journal of Traffic and Transportation Engineering (English Edition), vol. 7, no. 4, pp. 447–466, 2020

    2020

  2. [2]

    Hybrid spatio-temporal graph convolutional network: Improving traffic prediction with navi- gation data,

    R. Dai, S. Xu, Q. Gu, C. Ji, and K. Liu, “Hybrid spatio-temporal graph convolutional network: Improving traffic prediction with navi- gation data,” inProceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), 2020, pp. 3074–3082

    2020

  3. [3]

    Deep multi-view spatial-temporal network for taxi demand prediction,

    H. Yao, F. Wu, J. Ke, X. Tang, Y . Jia, S. Lu, P. Gong, J. Ye, and Z. Li, “Deep multi-view spatial-temporal network for taxi demand prediction,” inProceedings of the Thirty-Second AAAI Conference on Artificial Intelligence, 2018, pp. 2588–2595

    2018

  4. [4]

    Largest: A benchmark dataset for large- scale traffic forecasting,

    X. Liu, Y . Xia, Y . Liang, J. Hu, Y . Wang, L. Bai, C. Huang, Z. Liu, B. Hooi, and R. Zimmermann, “Largest: A benchmark dataset for large- scale traffic forecasting,” inAdvances in Neural Information Processing Systems (NeurIPS), Datasets and Benchmarks Track, vol. 36, 2023

    2023

  5. [5]

    Diffusion convolutional recurrent neural network: Data-driven traffic forecasting,

    Y . Li, R. Yu, C. Shahabi, and Y . Liu, “Diffusion convolutional recurrent neural network: Data-driven traffic forecasting,” inInternational Con- ference on Learning Representations (ICLR), 2018

    2018

  6. [6]

    Deep spatio-temporal residual networks for citywide crowd flows prediction,

    J. Zhang, Y . Zheng, and D. Qi, “Deep spatio-temporal residual networks for citywide crowd flows prediction,” inProceedings of the Thirty-First AAAI Conference on Artificial Intelligence, 2017, pp. 1655–1661

    2017

  7. [7]

    Spatio-temporal graph convolutional net- works: A deep learning framework for traffic forecasting,

    B. Yu, H. Yin, and Z. Zhu, “Spatio-temporal graph convolutional net- works: A deep learning framework for traffic forecasting,” inProceed- ings of the 27th International Joint Conference on Artificial Intelligence (IJCAI), 2018, pp. 3634–3640

    2018

  8. [8]

    Pdformer: Propagation delay-aware dynamic long-range transformer for traffic flow prediction,

    J. Jiang, C. Han, W. X. Zhao, and J. Wang, “Pdformer: Propagation delay-aware dynamic long-range transformer for traffic flow prediction,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 4, 2023, pp. 4365–4373

    2023

  9. [9]

    Decoupled dynamic spatial-temporal graph neural network for traffic forecasting,

    Z. Shao, Z. Zhang, W. Wei, F. Wang, Y . Xu, X. Cao, and C. S. Jensen, “Decoupled dynamic spatial-temporal graph neural network for traffic forecasting,”Proceedings of the VLDB Endowment, vol. 15, no. 11, pp. 2733–2746, 2022

    2022

  10. [10]

    Autostg: Neural architecture search for predictions of spatio-temporal graphs,

    Z. Pan, S. Ke, X. Yang, Y . Liang, Y . Yu, J. Zhang, and Y . Zheng, “Autostg: Neural architecture search for predictions of spatio-temporal graphs,” inProceedings of the Web Conference 2021 (WWW ’21), 2021

    2021

  11. [11]

    Autocts: Automated correlated time series forecasting,

    X. Wu, D. Zhang, C. Guo, C. He, B. Yang, and C. S. Jensen, “Autocts: Automated correlated time series forecasting,”Proceedings of the VLDB Endowment, vol. 15, no. 4, pp. 971–983, 2022

    2022

  12. [12]

    Autostf: Decoupled neural archi- tecture search for cost-effective automated spatio-temporal forecasting,

    T. Lyu, W. Zhang, J. Deng, and H. Liu, “Autostf: Decoupled neural archi- tecture search for cost-effective automated spatio-temporal forecasting,” inProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’25), 2025, pp. 985–996

    2025

  13. [13]

    Darts: Differentiable architec- ture search,

    H. Liu, K. Simonyan, and Y . Yang, “Darts: Differentiable architec- ture search,” inInternational Conference on Learning Representations (ICLR), 2019

    2019

  14. [14]

    Autost: Efficient neural architecture search for spatio-temporal prediction,

    T. Li, J. Zhang, K. Bao, Y . Liang, Y . Li, and Y . Zheng, “Autost: Efficient neural architecture search for spatio-temporal prediction,” in Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), 2020

    2020

  15. [15]

    Autostg+: An automatic framework to discover the optimal network for spatio- temporal graph prediction,

    S. Ke, Z. Pan, T. He, Y . Liang, J. Zhang, and Y . Zheng, “Autostg+: An automatic framework to discover the optimal network for spatio- temporal graph prediction,”Artificial Intelligence, vol. 318, p. 103899, 2023

    2023

  16. [16]

    Autostl: Automated spatio-temporal multi-task learning,

    Z. Zhang, X. Zhao, H. Miao, C. Zhang, H. Zhao, and J. Zhang, “Autostl: Automated spatio-temporal multi-task learning,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 37, 2023, pp. 4902– 4910

    2023

  17. [17]

    Traffic spatial-temporal prediction based on neural architecture search,

    D. Zhang, G. Luo, and J. Li, “Traffic spatial-temporal prediction based on neural architecture search,” inProceedings of the 18th International Symposium on Spatial and Temporal Data (SSTD ’23), 2023, pp. 21–30

    2023

  18. [18]

    Evolutionary neural architecture search for traffic forecasting,

    D. Klosa and C. B ¨uskens, “Evolutionary neural architecture search for traffic forecasting,” in2022 21st IEEE International Conference on Machine Learning and Applications (ICMLA), 2022, pp. 1230–1237

    2022

  19. [19]

    Low cost evolutionary neural architecture search (LENAS) applied to traffic forecasting,

    D. Klosa and C. B ¨uskens, “Low cost evolutionary neural architecture search (LENAS) applied to traffic forecasting,”Machine Learning and Knowledge Extraction, vol. 5, no. 3, pp. 830–846, 2023

    2023

  20. [20]

    Au- tocts+: Joint neural architecture and hyperparameter search for correlated time series forecasting,

    X. Wu, D. Zhang, M. Zhang, C. Guo, B. Yang, and C. S. Jensen, “Au- tocts+: Joint neural architecture and hyperparameter search for correlated time series forecasting,”Proceedings of the ACM on Management of Data, vol. 1, no. 1, pp. 97:1–97:26, 2023

    2023

  21. [21]

    Neural architecture search: A survey,

    T. Elsken, J. H. Metzen, and F. Hutter, “Neural architecture search: A survey,”Journal of Machine Learning Research, vol. 20, pp. 1–21, 2019

    2019

  22. [22]

    Neural architecture search: Insights from 1000 papers,

    C. White, M. Safari, R. Sukthanker, B. Ru, T. Elsken, A. Zela, D. Dey, and F. Hutter, “Neural architecture search: Insights from 1000 papers,” arXiv preprint arXiv:2301.08727, 2023

    Pith/arXiv arXiv 2023

  23. [23]

    Graph neural network for traffic forecasting: A survey,

    W. Jiang and J. Luo, “Graph neural network for traffic forecasting: A survey,”Expert Systems with Applications, vol. 207, p. 117921, 2022

    2022

  24. [24]

    Spatio-temporal graph neural networks for predictive learning in urban computing: A survey,

    G. Jin, Y . Liang, Y . Fang, Z. Shao, J. Huang, J. Zhang, and Y . Zheng, “Spatio-temporal graph neural networks for predictive learning in urban computing: A survey,”IEEE Transactions on Knowledge and Data Engineering, vol. 36, no. 10, pp. 5388–5407, 2024

    2024

  25. [25]

    A comprehensive review of traffic prediction: From traditional machine learning to automl,

    M. Khatiriolyaee, L. Yang, and R. W. Pazzi, “A comprehensive review of traffic prediction: From traditional machine learning to automl,” Neurocomputing, vol. 666, p. 132148, 2026

    2026

  26. [26]

    Spatio-temporal self-supervised learning for traffic flow prediction,

    J. Ji, J. Wang, C. Huang, J. Wu, B. Xu, Z. Wu, J. Zhang, and Y . Zheng, “Spatio-temporal self-supervised learning for traffic flow prediction,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 4, 2023, pp. 4356–4364

    2023

  27. [27]

    Multi-modality spatio- temporal forecasting via self-supervised learning,

    J. Deng, R. Jiang, J. Zhang, and X. Song, “Multi-modality spatio- temporal forecasting via self-supervised learning,” inProceedings of the 33rd International Joint Conference on Artificial Intelligence (IJCAI), 2024, pp. 2013–2021

    2024

  28. [28]

    Efficient large-scale traffic forecasting with transformers: A spatial data management perspective,

    Y . Fang, Y . Liang, B. Hui, Z. Shao, L. Deng, X. Liu, X. Jiang, and K. Zheng, “Efficient large-scale traffic forecasting with transformers: A spatial data management perspective,” inProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2025

    2025

  29. [29]

    Sagdfn: A scalable adaptive graph diffusion forecasting network for multivariate time series forecasting,

    Y . Jiang, X. Li, Y . Chen, S. Liu, W. Kong, A. F. Lentzakis, and G. Cong, “Sagdfn: A scalable adaptive graph diffusion forecasting network for multivariate time series forecasting,” inProceedings of the 40th IEEE International Conference on Data Engineering (ICDE), 2024

    2024

  30. [30]

    Traffic prediction with transfer learning: A mutual information-based approach,

    Y . Huang, X. Song, Y . Zhu, S. Zhang, and J. J. Yu, “Traffic prediction with transfer learning: A mutual information-based approach,”IEEE Transactions on Intelligent Transportation Systems, vol. 24, no. 8, pp. 8526–8540, 2023

    2023

  31. [31]

    Personalized federated learning for cross-city traffic prediction,

    Y . Zhang, H. Lu, N. Liu, Y . Xu, Q. Li, and L. Cui, “Personalized federated learning for cross-city traffic prediction,” inProceedings of the 33rd International Joint Conference on Artificial Intelligence (IJCAI), 2024, pp. 5526–5534

    2024

  32. [32]

    Diffstg: Probabilistic spatio-temporal graph forecasting with denoising diffusion models,

    H. Wen, Y . Lin, Y . Xia, H. Wan, Q. Wen, R. Zimmermann, and Y . Liang, “Diffstg: Probabilistic spatio-temporal graph forecasting with denoising diffusion models,” inProceedings of the 31st ACM International Con- ference on Advances in Geographic Information Systems (SIGSPATIAL), 2023

    2023

  33. [33]

    Unist: A prompt-empowered universal model for urban spatio-temporal prediction,

    Y . Yuan, J. Ding, J. Feng, D. Jin, and Y . Li, “Unist: A prompt-empowered universal model for urban spatio-temporal prediction,” inProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2024, pp. 4095–4106

    2024

  34. [34]

    Opencity: Open spatio-temporal foundation models for traffic prediction,

    Z. Li, L. Xia, L. Shi, Y . Xu, D. Yin, and C. Huang, “Opencity: Open spatio-temporal foundation models for traffic prediction,”ACM Transactions on Intelligent Systems and Technology, 2024

    2024

  35. [35]

    Advances in neural architecture search,

    X. Wang and W. Zhu, “Advances in neural architecture search,”National Science Review, vol. 11, no. 8, p. nwae282, 2024

    2024

  36. [36]

    Iot data analytics in dynamic environments: From an automated machine learning perspective,

    L. Yang and A. Shami, “Iot data analytics in dynamic environments: From an automated machine learning perspective,”Engineering Appli- cations of Artificial Intelligence, vol. 116, p. 105366, 2022

    2022

  37. [37]

    Optimizing time series forecasting architec- tures: A hierarchical neural architecture search approach,

    D. Deng and M. Lindauer, “Optimizing time series forecasting architec- tures: A hierarchical neural architecture search approach,”Transactions on Machine Learning Research, 2025

    2025

  38. [38]

    Neural architecture search with reinforcement learning,

    B. Zoph and Q. V . Le, “Neural architecture search with reinforcement learning,” inInternational Conference on Learning Representations (ICLR), 2017

    2017

  39. [39]

    Differentiable archi- tecture search with random features,

    X. Zhang, Y . Li, X. Zhang, Y . Wang, and J. Sun, “Differentiable archi- tecture search with random features,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023

    2023

  40. [40]

    Zero-cost operation scoring in differentiable architecture search,

    L. Xiang, Ł. Dudziak, M. S. Abdelfattah, T. Chau, N. D. Lane, and H. Wen, “Zero-cost operation scoring in differentiable architecture search,” inProceedings of the AAAI Conference on Artificial Intelli- gence, vol. 37, no. 9, 2023, pp. 10 577–10 585

    2023