Pith. sign in

REVIEW 3 major objections 5 minor 95 references

Embracing Large Language Models in Traffic Flow Forecasting

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A frozen large language model can improve traffic forecasting by choosing among candidate forecasts rather than generating them.

desk verdict A genuinely new LLM-as-selector mechanism with reproducible code, but the missing non-LLM selector control leaves the paper's central claim unproven. read the letter →

arxiv 2412.12201 v2 pith:3J6HQ6CG submitted 2024-12-15 cs.LG cs.AI

Yusheng Zhao , Xiao Luo , Haomin Wen , Zhiping Xiao , Wei Ju , Ming Zhang This is my paper · ORCID
classification cs.LGcs.AI
keywords trafficflowforecastinglargelanguagemodelstest-timeadaptationgraphneuralnetworkshypergraphrankinglossdistributionshiftspatio-temporal
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

The paper proposes LEAF, a traffic flow forecasting framework that uses a frozen large language model as a test-time selector rather than as a direct forecaster. Two learned branches, one graph-based and one hypergraph-based, each produce forecasts; the LLM reads a text prompt with historical data and candidate predictions, then picks the most plausible option for each sensor. The chosen options are fed back through a ranking loss to refine the two branches. On three public traffic datasets with only 10% of data used for training, LEAF reports consistent improvements over eight baselines, including lower MAE on every dataset. The central claim is that the discriminative ability of an LLM, not its generative ability, is what helps forecasting under distribution shift.

What carries the argument

The load-bearing mechanism is the prediction-selection loop built from a dual-branch predictor and an LLM-based selector. The graph branch uses graph convolutions over a spatio-temporal graph to model pair-wise relations; the hypergraph branch uses a low-rank learnable incidence matrix to model non-pair-wise relations. At test time, each branch emits a forecast, a small set of transformations (smoothing, upward/downward trend, over/under-estimation) expands these into a choice set per sensor, and a frozen LLM selects one option from a structured prompt. The selected option becomes a pseudo-label for a hinge-style ranking loss that pushes the predictor's output closer to the chosen candidate than to any rejected candidate; this loss is applied for a few iterations before the next prediction-selection round, with K=2 rounds in the reported experiments.

What would settle it

Replace the LLM selector on PEMS08 with a random pick from the same choice set, or with a fixed heuristic such as always choosing the downward-trend option during evening rush hours; if the resulting MAE matches or closely approaches LEAF's 24.68, the reported gains come from the choice set and ranking loop rather than from the LLM's reasoning.

Watch

Extended reading notes

Core claim

LEAF claims that a frozen LLM can serve as a reliable test-time selector for traffic flow forecasting, and that this selector-plus-ranking-loss loop improves prediction accuracy beyond either branch alone or any of eight baselines. For example, on PEMS08 the method achieves MAE 24.68 versus 26.42 for the best baseline, and the ablation shows that removing either branch, removing the transformations that expand the choice set, or removing the ranking loss all degrade performance. The paper interprets this as evidence that the LLM uses its internal knowledge of traffic patterns and rush-hour dynamics to pick the most likely candidate, and that supervising the predictor with these picks through a ranking loss yields further gains.

Load-bearing premise

The method assumes the frozen LLM's chosen candidate is a trustworthy pseudo-label for updating the predictor through the ranking loss, because the final output is the LLM's pick rather than the predictor's own forecast.

Editorial extensions

If this is right

  • LLMs do not need to generate numeric forecasts to help traffic prediction; selecting among candidate forecasts is a lower-risk use of their knowledge.
  • Combining pair-wise graph relations and non-pair-wise hypergraph relations in one predictor is beneficial, since ablations removing either branch degrade performance.
  • Test-time selection with ranking-loss supervision reduces long-horizon forecasting errors more than short-horizon errors, as shown by per-timestep MAE curves.
  • The framework works with small training sets (10% of data), suggesting it is useful in settings where labeled traffic data are scarce and distribution shift is expected.

Reading between the lines

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

  • A testable extension the paper does not run is to replace the LLM selector with a random pick or a simple heuristic rule (for example, always choose the downward-trend option during the evening rush) over the same choice set; matching LEAF's MAE would indicate the gains come from the choice set and ranking loop rather than from LLM reasoning.
  • Because the LLM is frozen and the prompt is purely textual, the same selector design should transfer to other spatio-temporal forecasting tasks with contextual text, such as energy load or crowd flow prediction, as long as a prompt can describe the location and time context.
  • The iterative loop doubles LLM inference cost when K=2, so deployment would likely benefit from prompt caching or batched selection across sensors; the paper does not report total LLM inference cost.
  • The ranking loss assumes the LLM's pick is at least better than the closest rejected candidate; if the LLM is wrong in a systematic way, the loop could reinforce that bias, which is why a heuristic baseline comparison matters.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. LEAF is a test-time traffic forecasting framework that combines a dual-branch predictor (a graph branch and a hypergraph branch) with a frozen LLM-based selector. The predictor is pretrained on a small training set; during inference, each branch produces per-vertex forecasts, and five fixed transformations (smoothing, upward/downward trend, overestimate/underestimate) expand these into a choice set. An LLM prompt containing historical data, spatio-temporal information, and the candidate sequences is used to select the most likely candidate per vertex, and the selected candidates are then treated as positives in a ranking loss that updates the predictor. This prediction-selection loop is repeated K times, and the final forecast is the LLM's selected candidates. Experiments on PEMS03, PEMS04, and PEMS08 report consistent improvements over eight baselines, with ablations on PEMS08 showing contributions from both branches, the transformations, and the ranking loss.

Significance. If the reported improvements are attributable to the proposed mechanism, the paper would provide a practical way to exploit the discriminative ability of frozen LLMs for traffic forecasting while keeping the LLM out of the generative loop, and it ships code and uses standard datasets and metrics. The idea of using LLM selections as ranking-loss supervision for a dual-branch predictor is interesting and goes beyond direct generative LLM forecasting. However, I agree with the stress-test concern that this attribution is currently unsupported: the final output is the LLM's pick, not the predictor's own output, and no cheap-selector control is run. The paper's significance would be substantially strengthened by such controls, by a clear validation protocol for M and K, and by an actual distribution-shift evaluation, which is the stated motivation of the work.

major comments (3)
  1. [§3.3 and Algorithm 1; Table 2] The final forecast is the LLM-selected candidate (Algorithm 1, lines 6-9), not the predictor's own output, and the ranking loss in Eqs. 7-8 is supervised by the same selection. The ablations E1-E6 show that adding the selector to a branch improves over that branch alone, but no control replaces the LLM with a random or simple heuristic selector (e.g., always choose the downward-trend hypergraph candidate when the forecast period is the end of rush hour, or choose the candidate whose trend best matches the historical slope). Without these controls, the reported gains could come entirely from the choice set and the prompt's domain hints rather than from the LLM's discriminative reasoning or from the predictor adaptation. Please add a random-selector baseline, two or three heuristic selectors, an oracle upper-bound selector, and an ablation where the final output is the predictor's own output after ranking-loss training.
  2. [§4.1, §4.4, Table 1] The hyperparameters M and K in Algorithm 1 are selected on PEMS08 (Figure 4), and Table 1 then reports PEMS08 as a test result without stating a validation split for that selection; the Table 2 ablations are also on PEMS08, so the headline PEMS08 improvement is at risk of selection bias. Furthermore, the paper's motivation is adaptation to test-time environmental changes (Abstract and §1), but the experiments only use a 10% training subset and non-overlapping test slices, which is not a controlled distribution-shift evaluation. Please report hyperparameter choices made on a validation split (or tuned on one dataset and transferred), provide error bars over multiple runs, and add at least one experiment with a genuine train/test distribution shift, such as different time periods or different weather conditions.
  3. [§4.2, Figure 8, Figure 6] The prompt in Figure 8 explicitly instructs the LLM that rush-hour phase is the most important temporal signal and explains how to infer the beginning or end of rush hours from historical changes. The example in Figure 6 shows the LLM selecting a downward-trend candidate because 'the rush hour is likely ending.' This injected domain knowledge is a legitimate part of the method, but it means a simple rule-based selector could reproduce much of its behavior. Please test the selector with prompts that omit these hints, or compare against a heuristic that encodes the same rush-hour rule, to support the claim that the LLM's internal knowledge and reasoning drive the improvements.
minor comments (5)
  1. [Table 1] The table header uses 'STSGNN' while §4.2 and the reference list use 'STSGCN'; please make the names consistent.
  2. [Eq. (3)] The expression for the normalized adjacency matrix, dAST = D^{-1/2}AST D^{-1/2}, is notationally unclear because the tilde or self-loop term is missing; please define the normalized matrix explicitly.
  3. [Figure 4] Both panels lack axis labels; please add the metric names and the hyperparameter values on the x-axes.
  4. [§4.1] The sentence 'We choose a subset of non-overlapping slices in the test set' should specify the subset size, the slice length, and the random seed for reproducibility.
  5. [Figure 8] The prompt contains the instruction 'Note that smoothing does not reduce MAE error'; if this is a deliberate design hint, it should be explained in the main text, otherwise remove it.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: LEAF's reported gains are measured against external PEMS test data, and the LLM selection loop is a test-time mechanism rather than a derivation that assumes its conclusion.

full rationale

The paper's derivation chain is empirical rather than definitional. The predictor branches (Eqs. 3-5) are standard GNN/hypergraph components; the choice set (Eq. 6) is constructed from the branches' outputs plus fixed transformations; the LLM selector picks a candidate; the ranking loss (Eqs. 7-8) updates the branches toward the selected candidate; and the final output (Algorithm 1, line 9) is the LLM's selected candidate. Every step is well-defined and the reported errors are computed against ground-truth PEMS03/04/08 data that are not used to construct the choice set or to prompt the LLM. No equation reduces to another by construction, and no fitted parameter is renamed as a prediction. The self-citations (e.g., Zhao et al. 2023 for the hypergraph low-rank incidence matrix) are to component designs, not to a uniqueness theorem or to the central claim that LLM selection improves accuracy. The absence of a random/heuristic selector control is a legitimate experimental-design concern about attribution of the gain, but it is not circularity: the final metric still measures an external quantity. The Limitations section candidly notes the LLM is not fine-tuned and only traffic data are considered, which does not indicate circularity.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The method's success depends on a handful of hand-set transformation magnitudes and loop lengths, plus the untested premise that an LLM's choices are trustworthy pseudo-labels. No formal derivation is involved, so the ledger below lists the empirical choices that carry the result.

free parameters (5)
  • trend slope range for upward/downward transformations = 1% to 12% linearly across 12 timesteps
    Chosen by hand in Section 4.1; controls how far the selector can deviate from base predictions. No sensitivity analysis for this range is reported.
  • overestimate/underestimate offset = ±5%
    Chosen by hand in Section 4.1; expands the choice set symmetrically. No ablation varies the offset magnitude.
  • M, ranking-loss update iterations = 5
    Set from hyperparameter analysis on PEMS08 (Figure 4a); if selected on test data, this is test-set tuning.
  • K, prediction-selection iterations = 2
    Set from hyperparameter analysis on PEMS08 (Figure 4b); the authors note performance degrades beyond K=2.
  • margin epsilon in ranking loss = 0
    Set in Section 4.1; a standard hinge-margin choice, not derived from data.
assumptions (4)
  • ad hoc to paper Linear transformations (smoothing, fixed offsets, linear trends) applied to base predictions produce a choice set that can contain a prediction close enough to the true future flow.
    Section 3.2 Eq. 6 and Section 4.1. If the true future deviates non-linearly from the base predictions, the selector cannot recover it; no experiment tests choice-set coverage.
  • domain assumption A frozen LLM's internal knowledge about traffic, rush hours, and time-of-day yields selections better than the base predictors or simple heuristics.
    Section 3.2 and Figure 3 prompt; no ablation compares LLM selection with random, oracle, or nearest-history heuristic selection.
  • domain assumption LLM-selected pseudo-labels are valid supervisory signals for the predictor in the absence of ground truth.
    Section 3.3 Eq. 7-8 and Algorithm 1 lines 6-7. If pseudo-labels are biased, ranking-loss adaptation encodes the bias.
  • standard math GCN and hypergraph propagation rules (Eq. 3-5) are appropriate for traffic data and are inherited from prior work.
    Equations 3-5 come from Kipf and Welling (2016) and Zhao et al. (2023); no new derivation is needed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Embracing Large Language Models in Traffic Flow Forecasting." pith.science (2026). https://pith.science/paper/3J6HQ6CG

@misc{pith2026241212201,
  author       = {Pith},
  title        = {Pith review of: Embracing Large Language Models in Traffic Flow Forecasting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3J6HQ6CG}},
  note         = {Machine review of arXiv:2412.12201}
}
read the original abstract

Traffic flow forecasting aims to predict future traffic flows based on the historical traffic conditions and the road network. It is an important problem in intelligent transportation systems, with a plethora of methods been proposed. Existing efforts mainly focus on capturing and utilizing spatio-temporal dependencies to predict future traffic flows. Though promising, they fall short in adapting to test-time environmental changes of traffic conditions. To tackle this challenge, we propose to introduce large language models (LLMs) to help traffic flow forecasting and design a novel method named Large Language Model Enhanced Traffic Flow Predictor (LEAF). LEAF adopts two branches, capturing different spatio-temporal relations using graph and hypergraph structures respectively. The two branches are first pre-trained individually, and during test-time, they yield different predictions. Based on these predictions, a large language model is used to select the most likely result. Then, a ranking loss is applied as the learning objective to enhance the prediction ability of the two branches. Extensive experiments on several datasets demonstrate the effectiveness of the proposed LEAF.

Figures

Figures reproduced from arXiv: 2412.12201 by the authors.

Figure 1
Figure 1. To use LLMs for traffic flow forecasting, [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The framework of the proposed LEAF, consisting of a dual-branch predictor and an LLM-based selector. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. An illustration of the prompt template. applied for normalization. Subsequently, the output features can be computed as: X (l) H = IH  I ⊤ HX (l−1) H + σ  WEI ⊤ HX (l−1) H  , (5) where WE ∈ R m×m models the interactions of the hyperedges. In this way, the hypergraph branch considers both (a) the interactions within a set of vertices (within a hyperedge) through the first term of Eq. 5, and (b) the interactions a… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: The forecasting errors under different hyper [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 6
Figure 6. Figure 6: The LLM’s analysis when selecting data. 1 2 3 4 5 6 7 8 9 10 11 12 Timestep 20 25 30 35 MAE Graph Branch Hypergraph Branch LEAF [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: The Mean Absolute Error (MAE) under dif [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 5
Figure 5. Figure 5: Visualization of historical data, choices, and [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 8
Figure 8. Figure 8: The details about the prompt [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

95 extracted references · 55 canonical work pages

  1. [1]

    AI@Meta. 2024. https://github.com/meta-llama/llama3/blob/main/MODEL_CARD.md Llama 3 model card

    2024

  2. [2]

    Abdul Fatir Ansari, Lorenzo Stella, Caner Turkmen, Xiyuan Zhang, Pedro Mercado, Huibin Shen, Oleksandr Shchur, Syama Sundar Rangapuram, Sebastian Pineda Arango, Shubham Kapoor, et al. 2024. Chronos: Learning the language of time series. arXiv preprint arXiv:2403.07815

    arXiv 2024

  3. [3]

    Shahrokh Asadi, Akbar Tavakoli, and Seyed Reza Hejazi. 2012. A new hybrid for improvement of auto-regressive integrated moving average models applying particle swarm optimization. Expert Systems with Applications, 39(5):5332--5337

    2012

  4. [4]

    Keqin Bao, Jizhi Zhang, Xinyu Lin, Yang Zhang, Wenjie Wang, and Fuli Feng. 2024. Large language models for recommendation: Past, present, and future. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 2993--2996

    2024

  5. [5]

    Himanshu Beniwal, Dishant Patel, D Kowsik, Hritik Ladia, Ankit Yadav, and Mayank Singh. 2024. Remember this event that year? assessing temporal information and understanding in large language models. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 16239--16348

    2024

  6. [6]

    Azzedine Boukerche, Yanjie Tao, and Peng Sun. 2020. Artificial intelligence-based vehicular traffic flow prediction methods for supporting intelligent transportation systems. Computer networks, 182:107484

    2020

  7. [7]

    Ling Cai, Krzysztof Janowicz, Gengchen Mai, Bo Yan, and Rui Zhu. 2020. Traffic transformer: Capturing the continuity and periodicity of time series for traffic forecasting. Transactions in GIS, 24(3):736--755

    2020

  8. [8]

    Yupeng Chang, Xu Wang, Jindong Wang, Yuan Wu, Linyi Yang, Kaijie Zhu, Hao Chen, Xiaoyuan Yi, Cunxiang Wang, Yidong Wang, et al. 2024. A survey on evaluation of large language models. ACM Transactions on Intelligent Systems and Technology, 15(3):1--45

    2024

Show all 95 references
  1. [9]

    Changlu Chen, Yanbin Liu, Ling Chen, and Chengqi Zhang. 2022 a . Bidirectional spatial-temporal adaptive transformer for urban traffic flow forecasting. IEEE Transactions on Neural Networks and Learning Systems, 34(10):6913--6925

    2022

  2. [10]

    Jian Chen, Li Zheng, Yuzhu Hu, Wei Wang, Hongxing Zhang, and Xiping Hu. 2024 a . Traffic flow matrix-based graph neural network with attention mechanism for traffic flow prediction. Information Fusion, 104:102146

    2024

  3. [11]

    Jin Chen, Zheng Liu, Xu Huang, Chenwang Wu, Qi Liu, Gangwei Jiang, Yuanhao Pu, Yuxuan Lei, Xiaolong Chen, Xingmei Wang, et al. 2024 b . When large language models meet personalization: Perspectives of challenges and opportunities. World Wide Web, 27(4):42

    2024

  4. [12]

    Ling Chen, Wei Shao, Mingqi Lv, Weiqi Chen, Youdong Zhang, and Chenghu Yang. 2022 b . Aargnn: An attentive attributed recurrent graph neural network for traffic flow prediction considering multiple dynamic factors. IEEE Transactions on Intelligent Transportation Systems, 23(10...

    2022

  5. [13]

    Mouxiang Chen, Lefei Shen, Han Fu, Zhuo Li, Jianling Sun, and Chenghao Liu. 2024 c . Calibration of time-series forecasting: Detecting and adapting context-driven distribution shift. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages...

    2024

  6. [14]

    Junyoung Chung, Caglar Gulcehre, KyungHyun Cho, and Yoshua Bengio. 2014. Empirical evaluation of gated recurrent neural networks on sequence modeling. arXiv preprint arXiv:1412.3555

    2014 arXiv

  7. [15]

    Fei Dai, Penggui Huang, Qi Mo, Xiaolong Xu, Muhammad Bilal, and Houbing Song. 2022. St-innet: Deep spatio-temporal inception networks for traffic flow prediction in smart cities. IEEE Transactions on Intelligent Transportation Systems, 23(10):19782--19794

    2022

  8. [16]

    George Dimitrakopoulos and Panagiotis Demestichas. 2010. Intelligent transportation systems. IEEE Vehicular Technology Magazine, 5(1):77--84

    2010

  9. [17]

    Wei Fan, Pengyang Wang, Dongkun Wang, Dongjie Wang, Yuanchun Zhou, and Yanjie Fu. 2023. Dish-ts: a general paradigm for alleviating distribution shift in time series forecasting. In Proceedings of the AAAI conference on artificial intelligence, volume 37, pages 7522--7529

    2023

  10. [18]

    Mohammadhani Fouladgar, Mostafa Parchami, Ramez Elmasri, and Amir Ghaderi. 2017. Scalable deep traffic flow neural networks for urban traffic congestion prediction. In 2017 international joint conference on neural networks (IJCNN), pages 2251--2258. IEEE

    2017

  11. [19]

    Bidisha Ghosh, Biswajit Basu, and Margaret O'Mahony. 2009. Multivariate short-term traffic flow forecasting using time-series analysis. IEEE transactions on intelligent transportation systems, 10(2):246--254

    2009

  12. [20]

    Nate Gruver, Marc Finzi, Shikai Qiu, and Andrew G Wilson. 2024. Large language models are zero-shot time series forecasters. Advances in Neural Information Processing Systems, 36

    2024

  13. [21]

    Shengnan Guo, Youfang Lin, Ning Feng, Chao Song, and Huaiyu Wan. 2019. Attention based spatial-temporal graph convolutional networks for traffic flow forecasting. In Proceedings of the AAAI conference on artificial intelligence, volume 33, pages 922--929

    2019

  14. [22]

    Xusen Guo, Qiming Zhang, Junyue Jiang, Mingxing Peng, Meixin Zhu, and Hao Frank Yang. 2024 a . Towards explainable traffic flow prediction with large language models. Communications in Transportation Research, 4:100150

    2024

  15. [23]

    Xusen Guo, Qiming Zhang, Mingxing Peng, Meixin Zhua, et al. 2024 b . Explainable traffic flow prediction with large language models. arXiv preprint arXiv:2404.02937

    2024

  16. [24]

    Will Hamilton, Zhitao Ying, and Jure Leskovec. 2017. Inductive representation learning on large graphs. Advances in neural information processing systems, 30

    2017

  17. [25]

    Xiao Han, Zhenduo Zhang, Yiling Wu, Xinfeng Zhang, and Zhe Wu. 2024. Event traffic forecasting with sparse multimodal data. In Proceedings of the 32nd ACM International Conference on Multimedia, pages 8855--8864

    2024

  18. [26]

    S Hochreiter. 1997. Long short-term memory. Neural Computation MIT-Press

    1997

  19. [27]

    Wei-Chiang Hong, Yucheng Dong, Feifeng Zheng, and Shih Yung Wei. 2011. Hybrid evolutionary algorithms in a svr traffic flow forecasting model. Applied Mathematics and Computation, 217(15):6733--6747

    2011

  20. [28]

    Qisheng Hu, Geonsik Moon, and Hwee Tou Ng. 2024. From moments to milestones: Incremental timeline summarization leveraging large language models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7232--7246

    2024

  21. [29]

    Chun-Yao Jiang, Xiao-Min Hu, and Wei-Neng Chen. 2021. An urban traffic signal control system based on traffic flow prediction. In 2021 13th international conference on advanced computational intelligence (ICACI), pages 259--265. IEEE

    2021

  22. [30]

    Jiawei Jiang, Chengkai Han, Wayne Xin Zhao, and Jingyuan Wang. 2023. Pdformer: Propagation delay-aware dynamic long-range transformer for traffic flow prediction. In Proceedings of the AAAI conference on artificial intelligence, volume 37, pages 4365--4373

    2023

  23. [31]

    Ming Jin, Shiyu Wang, Lintao Ma, Zhixuan Chu, James Y Zhang, Xiaoming Shi, Pin-Yu Chen, Yuxuan Liang, Yuan-Fang Li, Shirui Pan, et al. 2023. Time-llm: Time series forecasting by reprogramming large language models. arXiv preprint arXiv:2310.01728

    2023 arXiv

  24. [32]

    Wei Ju, Yusheng Zhao, Yifang Qin, Siyu Yi, Jingyang Yuan, Zhiping Xiao, Xiao Luo, Xiting Yan, and Ming Zhang. 2024. Cool: a conjoint perspective on spatio-temporal graph neural network for traffic forecasting. Information Fusion, 107:102341

    2024

  25. [33]

    Shafiza Ariffin Kashinath, Salama A Mostafa, Aida Mustapha, Hairulnizam Mahdin, David Lim, Moamin A Mahmoud, Mazin Abed Mohammed, Bander Ali Saleh Al-Rimy, Mohd Farhan Md Fudzee, and Tan Jhon Yang. 2021. Review of data fusion methods for real-time and multi-sensor traffic flow...

    2021

  26. [34]

    Taesung Kim, Jinhee Kim, Yunwon Tae, Cheonbok Park, Jang-Ho Choi, and Jaegul Choo. 2021. Reversible instance normalization for accurate time-series forecasting against distribution shift. In International Conference on Learning Representations

    2021

  27. [35]

    Thomas N Kipf and Max Welling. 2016. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907

    2016 arXiv

  28. [36]

    Ross Koval, Nicholas Andrews, and Xifeng Yan. 2024. Financial forecasting from textual and tabular time series. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 8289--8300

    2024

  29. [37]

    Gonzalez, Hao Zhang, and Ion Stoica

    Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph E. Gonzalez, Hao Zhang, and Ion Stoica. 2023. Efficient memory management for large language model serving with pagedattention. In Proceedings of the ACM SIGOPS 29th Symposium on Operating S...

    2023

  30. [38]

    Yanis Labrak, Adrien Bazoge, Emmanuel Morin, Pierre-Antoine Gourraud, Mickael Rouvier, and Richard Dufour. 2024. Biomistral: A collection of open-source pretrained large language models for medical domains. arXiv preprint arXiv:2402.10373

    2024 arXiv

  31. [39]

    Jinqi Lai, Wensheng Gan, Jiayang Wu, Zhenlian Qi, and S Yu Philip. 2024. Large language models in law: A survey. AI Open

    2024

  32. [40]

    Shiyong Lan, Yitong Ma, Weikang Huang, Wenwu Wang, Hongyu Yang, and Pyang Li. 2022. Dstagnn: Dynamic spatial-temporal aware graph neural network for traffic flow forecasting. In International conference on machine learning, pages 11906--11917. PMLR

    2022

  33. [41]

    Fuxian Li, Jie Feng, Huan Yan, Guangyin Jin, Fan Yang, Funing Sun, Depeng Jin, and Yong Li. 2023. Dynamic graph convolutional recurrent network for traffic prediction: Benchmark and solution. ACM Transactions on Knowledge Discovery from Data, 17(1):1--21

    2023

  34. [42]

    Hourun Li, Yusheng Zhao, Zhengyang Mao, Yifang Qin, Zhiping Xiao, Jiaqi Feng, Yiyang Gu, Wei Ju, Xiao Luo, and Ming Zhang. 2024 a . Graph neural networks in intelligent transportation systems: Advances, applications and trends. arXiv preprint arXiv:2401.0713

    2024

  35. [43]

    Mengzhang Li and Zhanxing Zhu. 2021. Spatial-temporal fusion graph neural networks for traffic flow forecasting. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pages 4189--4196

    2021

  36. [44]

    Nian Li, Chen Gao, Mingyu Li, Yong Li, and Qingmin Liao. 2024 b . Econagent: large language model-empowered agents for simulating macroeconomic activities. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages...

    2024

  37. [45]

    Shuangshuang Li, Zhen Shen, and Gang Xiong. 2012. A k-nearest neighbor locally weighted regression method for short-term traffic flow forecasting. In 2012 15th International IEEE Conference on Intelligent Transportation Systems, pages 1596--1601. IEEE

    2012

  38. [46]

    Yaguang Li, Rose Yu, Cyrus Shahabi, and Yan Liu. 2018. Diffusion convolutional recurrent neural network: Data-driven traffic forecasting. In International Conference on Learning Representations

    2018

  39. [47]

    Zhonghang Li, Lianghao Xia, Jiabin Tang, Yong Xu, Lei Shi, Long Xia, Dawei Yin, and Chao Huang. 2024 c . Urbangpt: Spatio-temporal large language models. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 5351--5362

    2024

  40. [48]

    Yuebing Liang, Yichao Liu, Xiaohan Wang, and Zhan Zhao. 2024. Exploring large language models for human mobility prediction under public events. Computers, Environment and Urban Systems, 112:102153

    2024

  41. [49]

    Thomas Liebig, Nico Piatkowski, Christian Bockermann, and Katharina Morik. 2017. Dynamic route planning with real-time traffic predictions. Information Systems, 64:258--265

    2017

  42. [50]

    Valentin Li \'e vin, Christoffer Egeberg Hother, Andreas Geert Motzfeldt, and Ole Winther. 2024. Can large language models reason about medical questions? Patterns, 5(3)

    2024

  43. [51]

    Chenxi Liu, Sun Yang, Qianxiong Xu, Zhishuai Li, Cheng Long, Ziyue Li, and Rui Zhao. 2024 a . Spatial-temporal large language model for traffic prediction. arXiv preprint arXiv:2401.10134

    2024 arXiv

  44. [52]

    Hangchen Liu, Zheng Dong, Renhe Jiang, Jiewen Deng, Jinliang Deng, Quanjun Chen, and Xuan Song. 2023. Spatio-temporal adaptive embedding makes vanilla transformer sota for traffic forecasting. In Proceedings of the 32nd ACM international conference on information and knowledge...

    2023

  45. [53]

    Yong Liu, Haoran Zhang, Chenyu Li, Xiangdong Huang, Jianmin Wang, and Mingsheng Long. 2024 b . Timer: Generative pre-trained transformers are large time series models. arXiv preprint arXiv:2402.02368

    2024 arXiv

  46. [54]

    Wang Lu, Jindong Wang, Xinwei Sun, Yiqiang Chen, and Xing Xie. 2022. Out-of-distribution representation learning for time series classification. arXiv preprint arXiv:2209.07027

    2022 arXiv

  47. [55]

    Junyu Luo, Weizhi Zhang, Ye Yuan, Yusheng Zhao, Junwei Yang, Yiyang Gu, Bohan Wu, Binqi Chen, Ziyue Qiao, Qingqing Long, et al. 2025. Large language model agent: A survey on methodology, applications and challenges. arXiv preprint arXiv:2503.21460

    2025 arXiv

  48. [56]

    Xiaoyi Luo, Jiaheng Peng, and Jun Liang. 2022. Directed hypergraph attention network for traffic forecasting. IET Intelligent Transport Systems, 16(1):85--98

    2022

  49. [57]

    Ying Ma, Haijie Lou, Ming Yan, Fanghui Sun, and Guoqi Li. 2024. Spatio-temporal fusion graph convolutional network for traffic flow forecasting. Information Fusion, 104:102196

    2024

  50. [58]

    Silvia Milano, Joshua A McGrane, and Sabina Leonelli. 2023. Large language models challenge the future of higher education. Nature Machine Intelligence, 5(4):333--334

    2023

  51. [59]

    Shervin Minaee, Tomas Mikolov, Narjes Nikzad, Meysam Chenaghlu, Richard Socher, Xavier Amatriain, and Jianfeng Gao. 2024. Large language models: A survey. arXiv preprint arXiv:2402.06196

    2024 arXiv

  52. [60]

    Yilong Ren, Yue Chen, Shuai Liu, Boyue Wang, Haiyang Yu, and Zhiyong Cui. 2024. Tpllm: A traffic prediction framework based on pretrained large language models. arXiv preprint arXiv:2403.02221

    2024 arXiv

  53. [61]

    Nimra Shahid, Munam Ali Shah, Abid Khan, Carsten Maple, and Gwanggil Jeon. 2021. Towards greener smart cities and road traffic forecasting using air pollution data. Sustainable Cities and Society, 72:103062

    2021

  54. [62]

    Brian L Smith and Michael J Demetsky. 1997. Traffic flow forecasting: comparison of modeling approaches. Journal of transportation engineering, 123(4):261--266

    1997

  55. [63]

    Kihyuk Sohn. 2016. Improved deep metric learning with multi-class n-pair loss objective. Advances in neural information processing systems, 29

    2016

  56. [64]

    Chao Song, Youfang Lin, Shengnan Guo, and Huaiyu Wan. 2020. Spatial-temporal synchronous graph convolutional networks: A new framework for spatial-temporal network data forecasting. In Proceedings of the AAAI conference on artificial intelligence, volume 34, pages 914--921

    2020

  57. [65]

    Shiliang Sun, Changshui Zhang, and Guoqiang Yu. 2006. A bayesian network approach to traffic flow forecasting. IEEE Transactions on intelligent transportation systems, 7(1):124--132

    2006

  58. [66]

    Yanfeng Sun, Xiangheng Jiang, Yongli Hu, Fuqing Duan, Kan Guo, Boyue Wang, Junbin Gao, and Baocai Yin. 2022. Dual dynamic spatial-temporal graph convolution network for traffic prediction. IEEE Transactions on Intelligent Transportation Systems, 23(12):23680--23693

    2022

  59. [67]

    Tigran T Tchrakian, Biswajit Basu, and Margaret O'Mahony. 2011. Real-time traffic flow forecasting using spectral analysis. IEEE Transactions on Intelligent Transportation Systems, 13(2):519--526

    2011

  60. [68]

    Dave Van Veen, Cara Van Uden, Louis Blankemeier, Jean-Benoit Delbrouck, Asad Aali, Christian Bluethgen, Anuj Pareek, Malgorzata Polacin, Eduardo Pontes Reis, Anna Seehofnerov \'a , et al. 2024. Adapted large language models can outperform medical experts in clinical text summa...

    2024

  61. [69]

    A Vaswani. 2017. Attention is all you need. Advances in Neural Information Processing Systems

    2017

  62. [70]

    Petar Veli c kovi \'c , Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Li \`o , and Yoshua Bengio. 2018. Graph attention networks. In International Conference on Learning Representations

    2018

  63. [71]

    Jingcheng Wang, Yong Zhang, Yongli Hu, and Baocai Yin. 2024. Large-scale traffic prediction with hierarchical hypergraph message passing networks. IEEE Transactions on Computational Social Systems

    2024

  64. [72]

    Jingcheng Wang, Yong Zhang, Lixun Wang, Yongli Hu, Xinglin Piao, and Baocai Yin. 2022. Multitask hypergraph convolutional networks: A heterogeneous traffic prediction framework. IEEE Transactions on Intelligent Transportation Systems, 23(10):18557--18567

    2022

  65. [73]

    Xiaoyang Wang, Yao Ma, Yiqi Wang, Wei Jin, Xin Wang, Jiliang Tang, Caiyan Jia, and Jian Yu. 2020. Traffic flow prediction via spatial temporal graph neural network. In Proceedings of the web conference 2020, pages 1082--1092

    2020

  66. [74]

    Yi Wang and Di Zhu. 2022. Shgcn: a hypergraph-based deep learning model for spatiotemporal traffic flow prediction. In Proceedings of the 5th ACM SIGSPATIAL International Workshop on AI for Geographic Knowledge Discovery, pages 30--39

    2022

  67. [75]

    Kilian Q Weinberger and Lawrence K Saul. 2009. Distance metric learning for large margin nearest neighbor classification. Journal of machine learning research, 10(2)

    2009

  68. [76]

    Wenchao Weng, Jin Fan, Huifeng Wu, Yujie Hu, Hao Tian, Fu Zhu, and Jia Wu. 2023. A decomposition dynamic graph convolutional recurrent network for traffic forecasting. Pattern Recognition, 142:109670

    2023

  69. [77]

    G Woo, C Liu, A Kumar, C Xiong, S Savarese, and D Sahoo. 2024. Unified training of universal time series forecasting transformers. arxiv 2024. arXiv preprint arXiv:2402.02592

    2024 arXiv

  70. [78]

    Yuwei Xia, Ding Wang, Qiang Liu, Liang Wang, Shu Wu, and Xiao-Yu Zhang. 2024. Chain-of-history reasoning for temporal knowledge graph forecasting. In Findings of the Association for Computational Linguistics ACL 2024, pages 16144--16159

    2024

  71. [79]

    Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. 2018. How powerful are graph neural networks? In International Conference on Learning Representations

    2018

  72. [80]

    Mingxing Xu, Wenrui Dai, Chunmiao Liu, Xing Gao, Weiyao Lin, Guo-Jun Qi, and Hongkai Xiong. 2020. Spatial-temporal transformer networks for traffic flow forecasting. arXiv preprint arXiv:2001.02908

    2020 arXiv

  73. [81]

    Hao-Fan Yang, Tharam S Dillon, and Yi-Ping Phoebe Chen. 2016. Optimized structure of the traffic flow forecasting model with a deep learning approach. IEEE transactions on neural networks and learning systems, 28(10):2371--2381

    2016

  74. [82]

    Shukang Yin, Chaoyou Fu, Sirui Zhao, Ke Li, Xing Sun, Tong Xu, and Enhong Chen. 2024. A survey on multimodal large language models. National Science Review, page nwae403

    2024

  75. [83]

    Bing Yu, Haoteng Yin, and Zhanxing Zhu. 2018. Spatio-temporal graph convolutional networks: a deep learning framework for traffic forecasting. In Proceedings of the 27th International Joint Conference on Artificial Intelligence, pages 3634--3640

    2018

  76. [84]

    Xinli Yu, Zheng Chen, and Yanbin Lu. 2023. Harnessing llms for temporal data-a study on explainable financial time series forecasting. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 739--753

    2023

  77. [85]

    Yuan Yuan, Jingtao Ding, Jie Feng, Depeng Jin, and Yong Li. 2024. Unist: a prompt-empowered universal model for urban spatio-temporal prediction. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 4095--4106

    2024

  78. [86]

    Junping Zhang, Fei-Yue Wang, Kunfeng Wang, Wei-Hua Lin, Xin Xu, and Cheng Chen. 2011. Data-driven intelligent transportation systems: A survey. IEEE Transactions on Intelligent Transportation Systems, 12(4):1624--1639

    2011

  79. [87]

    Xiyue Zhang, Chao Huang, Yong Xu, Lianghao Xia, Peng Dai, Liefeng Bo, Junbo Zhang, and Yu Zheng. 2021. Traffic flow forecasting with spatial-temporal graph diffusion network. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pages 15008--15015

    2021

  80. [88]

    Zhiqiang Zhang, Dandan Zhang, and Yun Wang. 2024. Fast long sequence time-series forecasting for edge service running state based on data drift and non-stationarity. IEEE Transactions on Knowledge and Data Engineering, 36(11):6181--6194

    2024

  81. [89]

    Yusheng Zhao, Junyu Luo, Xiao Luo, Jinsheng Huang, Jingyang Yuan, Zhiping Xiao, and Ming Zhang. 2025 a . Attention bootstrapping for multi-modal test-time adaptation. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 22849--22857

    2025

  82. [90]

    Yusheng Zhao, Xiao Luo, Wei Ju, Chong Chen, Xian-Sheng Hua, and Ming Zhang. 2023. Dynamic hypergraph structure learning for traffic flow forecasting. In 2023 IEEE 39th International Conference on Data Engineering (ICDE), pages 2303--2316. IEEE

    2023

  83. [91]

    Yu, and Ming Zhang

    Yusheng Zhao, Xiao Luo, Weizhi Zhang, Wei Ju, Zhiping Xiao, Philip S. Yu, and Ming Zhang. 2025 b . Marco: Meta-reflection with cross-referencing for code reasoning. arXiv preprint arXiv:2505.17481

    2025 arXiv

  84. [92]

    Yusheng Zhao, Changhu Wang, Xiao Luo, Junyu Luo, Wei Ju, Zhiping Xiao, and Ming Zhang. 2025 c . Traci: A data-centric approach for multi-domain generalization on graphs. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 13401--13409

    2025

  85. [93]

    Chuanpan Zheng, Xiaoliang Fan, Shirui Pan, Haibing Jin, Zhaopeng Peng, Zonghan Wu, Cheng Wang, and S Yu Philip. 2023 a . Spatio-temporal joint graph convolutional networks for traffic forecasting. IEEE Transactions on Knowledge and Data Engineering, 36(1):372--385

    2023

  86. [94]

    Ou Zheng, Mohamed Abdel-Aty, Dongdong Wang, Zijin Wang, and Shengxuan Ding. 2023 b . Chatgpt is on the horizon: could a large language model be suitable for intelligent traffic safety research and applications? arXiv preprint arXiv:2303.05382

    2023 arXiv

  87. [95]

    Dongcheng Zou, Senzhang Wang, Xuefeng Li, Hao Peng, Yuandong Wang, Chunyang Liu, Kehua Sheng, and Bo Zhang. 2024. Multispans: A multi-range spatial-temporal transformer network for traffic forecast via structural entropy optimization. In Proceedings of the 17th ACM Internation...

    2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.