pith. sign in

arxiv: 2605.29768 · v2 · pith:S32HQEGVnew · submitted 2026-05-28 · 💻 cs.AI

From XXLTraffic to EvoXXLTraffic: Scaling Traffic Forecasting to Sensor-Evolving Networks

Du Yin , Hao Xue , Arian Prabowo , Shuang Ao , Flora Salim This is my paper

Pith reviewed 2026-06-29 07:33 UTC · model grok-4.3

classification 💻 cs.AI
keywords traffic forecastingevolving sensor networksspatio-temporal GNNcontinual learningdatasetPeMSroad networksstreaming forecasting
0
0 comments X

The pith

Traffic sensor networks that grow over decades make many state-of-the-art forecasting models ineffective.

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

The paper establishes that traffic forecasting benchmarks must account for sensor networks that expand as new sensors are installed over years. Existing fixed-sensor datasets do not capture this growth, which occurs at rates from hundreds to thousands of percent. By creating the EvoXXLTraffic dataset with yearly graph snapshots and a streaming protocol, the work shows that models successful on static benchmarks lose performance in this setting. A sympathetic reader would care because deployed systems in real cities face continuous network changes and cannot rely on outdated accuracy claims. This shifts focus from static spatio-temporal models to those handling evolution.

Core claim

Existing traffic forecasting assumes a fixed sensor set, but real networks grow continuously. The EvoXXLTraffic dataset reorganizes PeMS and Transport for NSW data into per-year active sensors, traffic-flow matrices, and graph snapshots spanning up to 27 years with growth ratios up to over 10,000%. Under a yearly streaming forecasting protocol, many state-of-the-art methods no longer achieve their reported results, better reflecting real-world conditions.

What carries the argument

The sensor-evolving reorganization of traffic data into yearly snapshots and the yearly streaming forecasting protocol on EvoXXLTraffic.

Load-bearing premise

Reorganizing the records into per-year active sensors accurately captures genuine network growth without introducing artifacts from labeling or cleaning choices.

What would settle it

Running the same baselines on a different city's traffic data with independently verified sensor addition dates and checking if the performance drop matches the paper's observations.

Figures

Figures reproduced from arXiv: 2605.29768 by Arian Prabowo, Du Yin, Flora Salim, Hao Xue, Shuang Ao.

Figure 1
Figure 1. Figure 1: Our dataset is evolving and longer than existing datasets. Existing datasets are either limited by short temporal spans or [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Sensor-network growth and adjacency snapshots across PeMS districts. Each panel shows the yearly active-sensor count [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comprehensive overview of the XXLTraffic dataset. (a) Global and regional sensor layouts. (b, c) Sensor traffic status distributions [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison of XXLTraffic gap forecasting and EvoXXLTraffic sensor-evolving forecasting protocols. [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Model training time comparison on the gap-forecasting subsets. The training time for all baselines per epoch is measured in [PITH_FULL_IMAGE:figures/full_fig_p022_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Training efficiency vs. MAE on PEMS04 (EvoXXLTraffic) at four prediction horizons. Each bubble is one baseline; [PITH_FULL_IMAGE:figures/full_fig_p023_6.png] view at source ↗
read the original abstract

Existing traffic forecasting benchmarks assume a fixed sensor set, but real road-sensor networks grow continuously as the road network changes year by year. We introduce the XXLTraffic dataset family, which spans up to 27 years of California PeMS and Transport for NSW data. The fixed-sensor subsets of XXLTraffic support extremely long forecasting with multi-year gaps and standard hourly / daily long-horizon forecasting. We extend it to EvoXXLTraffic, a sensor-evolving reorganization that exposes per-year active sensors, yearly traffic-flow matrices, and yearly graph snapshots across nine PeMS districts, with growth ratios ranging from +305% to over +10,000%. We define a yearly streaming forecasting protocol on EvoXXLTraffic in which each calendar year is a continual task, and benchmark a wide range of representative baselines drawn from static spatio-temporal GNNs, na\"ive online schemes, evolving-graph continual methods, and retrieval / test-time methods. We find that our ultra-large evolutionary dataset better reflects the real world, and many state-of-the-art (SOTA) results no longer work. Our dataset complements existing benchmarks by enabling more realistic forecasting under ultra-long evolutionary road networks. Our code and baselines are available at github repo: https://github.com/cruiseresearchgroup/TSAS26-EvoXXLTraffic

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces the XXLTraffic dataset family from up to 27 years of California PeMS and Transport for NSW traffic records. Fixed-sensor subsets enable long-horizon forecasting with multi-year gaps. EvoXXLTraffic reorganizes the data into sensor-evolving yearly snapshots exposing per-year active sensors, traffic-flow matrices, and graph snapshots with growth ratios from +305% to over +10,000%. A yearly streaming forecasting protocol is defined, and baselines spanning static spatio-temporal GNNs, naïve online schemes, evolving-graph continual methods, and retrieval/test-time approaches are benchmarked. The authors conclude that the evolutionary dataset better reflects the real world and that many SOTA results no longer hold.

Significance. If the reorganization faithfully isolates genuine network growth, the work supplies a large-scale benchmark that directly challenges the fixed-sensor assumption prevalent in traffic forecasting. This could drive development of continual and evolving-graph methods. Public release of code and baselines is a clear strength supporting reproducibility.

major comments (2)
  1. [Abstract] Abstract: The central claim that EvoXXLTraffic 'better reflects the real world' and that 'many state-of-the-art (SOTA) results no longer work' is load-bearing on the fidelity of the sensor-evolving reorganization. The description of 'per-year active sensors' and 'yearly graph snapshots' supplies no criteria for determining sensor activation dates, overlap statistics, or mitigation of upstream cleaning/relabeling artifacts common in long-term archives; performance drops could therefore arise from labeling inconsistencies rather than evolutionary dynamics.
  2. [Abstract] Abstract: Growth ratios (+305% to >+10,000%) are reported without accompanying per-year sensor counts, data-completeness metrics, or external validation against deployment records. This omission prevents assessment of whether the yearly snapshots isolate true network expansion or embed retrospective labeling choices.
minor comments (2)
  1. The escaped quote in 'na"ive' should be rendered as 'naive'.
  2. The GitHub repository link is given but the manuscript would benefit from a one-sentence summary of its contents (implemented baselines, data loaders, etc.).

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive comments on the fidelity of the EvoXXLTraffic reorganization. We address each point below and will revise the manuscript accordingly to improve transparency.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that EvoXXLTraffic 'better reflects the real world' and that 'many state-of-the-art (SOTA) results no longer work' is load-bearing on the fidelity of the sensor-evolving reorganization. The description of 'per-year active sensors' and 'yearly graph snapshots' supplies no criteria for determining sensor activation dates, overlap statistics, or mitigation of upstream cleaning/relabeling artifacts common in long-term archives; performance drops could therefore arise from labeling inconsistencies rather than evolutionary dynamics.

    Authors: We agree that the abstract omits key construction details. In the revised manuscript we will add an explicit subsection describing the activation criterion (a sensor is marked active in year Y if it contributes at least one valid hourly reading in the raw PeMS files for that calendar year), a table of year-to-year overlap percentages, and a short discussion of how we avoided additional relabeling by using the original district-level archives. These additions will allow readers to evaluate whether observed performance changes stem from network growth rather than labeling artifacts. revision: yes

  2. Referee: [Abstract] Abstract: Growth ratios (+305% to >+10,000%) are reported without accompanying per-year sensor counts, data-completeness metrics, or external validation against deployment records. This omission prevents assessment of whether the yearly snapshots isolate true network expansion or embed retrospective labeling choices.

    Authors: We will insert a new table (or expand Table 1) listing, for each district and year, the exact number of active sensors, the fraction of hours with valid readings, and the growth ratio computed directly from those counts. External validation against official deployment logs is not feasible because the public PeMS releases do not include linked deployment-date metadata; we will explicitly note this limitation and its implications for interpreting the growth figures. revision: partial

standing simulated objections not resolved
  • External validation of sensor activation dates against official deployment records (not available in the public data sources used)

Circularity Check

0 steps flagged

No circularity: empirical dataset paper with no derivations or self-referential fitting.

full rationale

The manuscript introduces XXLTraffic and EvoXXLTraffic via reorganization of public PeMS/NSW archives into yearly snapshots, then reports empirical benchmarks of existing methods. No equations, no fitted parameters renamed as predictions, and no load-bearing self-citations or uniqueness theorems appear in the provided text. The central claim (SOTA methods fail on the new evolutionary setting) rests on external comparisons to published baselines rather than any reduction to the authors' own prior definitions or fits. This matches the default expectation of a self-contained empirical contribution.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Dataset contribution paper; no free parameters or invented physical entities are introduced. The central claim rests on the domain assumption that the chosen public traffic records faithfully represent sensor network evolution.

axioms (1)
  • domain assumption Public traffic records from PeMS and Transport for NSW can be reorganized into per-year active sensor sets and graph snapshots that reflect genuine network growth.
    Invoked when constructing EvoXXLTraffic from the raw data sources.

pith-pipeline@v0.9.1-grok · 5775 in / 1209 out tokens · 26242 ms · 2026-06-29T07:33:38.867771+00:00 · methodology

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

58 extracted references · 7 canonical work pages · 1 internal anchor

  1. [1]

    Lei Bai, Lina Yao, Can Li, Xianzhi Wang, and Can Wang. 2020. Adaptive graph convolutional recurrent network for traffic forecasting.Advances in neural information processing systems33 (2020), 17804–17815

    2020

  2. [2]

    Chao Chen, Karl Petty, Alexander Skabardonis, Pravin Varaiya, and Zhanfeng Jia. 2001. Freeway performance measurement system: mining loop detector data.Transportation research record1748 (2001), 96–102

    2001

  3. [3]

    Wei Chen and Yuxuan Liang. 2025. Expand and compress: Exploring tuning principles for continual spatio-temporal graph forecasting. InInternational Conference on Learning Representations, Vol. 2025. 81631–81656

    2025

  4. [4]

    Wei Chen and Yuxuan Liang. 2025. Learning with calibration: Exploring test-time computing of spatio-temporal forecasting.Advances in Neural Information Processing Systems38 (2025), 155895–155929

    2025

  5. [5]

    Xinyu Chen and Lijun Sun. 2021. Bayesian temporal factorization for multidimensional time series prediction.IEEE Transactions on Pattern Analysis and Machine Intelligence44 (2021), 4659–4673

    2021

  6. [6]

    Xu Chen, Junshan Wang, and Kunqing Xie. 2021. TrafficStream: A Streaming Traffic Flow Forecasting Framework Based on Graph Neural Networks and Continual Learning. InProceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI-21, Zhi-Hua Zhou (Ed.). International Joint Conferences on Artificial Intelligence Organization, 3...

    work page doi:10.24963/ijcai.2021/498 2021

  7. [7]

    Xu Chen, Junshan Wang, and Kunqing Xie. 2021. TrafficStream: A streaming traffic flow forecasting framework based on graph neural networks and continual learning.arXiv preprint arXiv:2106.06273(2021)

    work page arXiv 2021

  8. [8]

    Razvan-Gabriel Cirstea, Chenjuan Guo, Bin Yang, Tung Kieu, Xuanyi Dong, and Shirui Pan. 2022. Triformer: Triangular, Variable-Specific Attentions for Long Sequence Multivariate Time Series Forecasting. InProceedings of the Thirty-First International Joint Conference on Artificial Intelligence, IJCAI-22, Lud De Raedt (Ed.). International Joint Conferences ...

    work page doi:10.24963/ijcai.2022/277 2022

  9. [9]

    Prathamesh Deshpande and Sunita Sarawagi. 2019. Streaming adaptation of deep forecasting models using adaptive recurrent units. InProceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 1560–1568

    2019

  10. [10]

    Zheng Fang, Qingqing Long, Guojie Song, and Kunqing Xie. 2021. Spatial-temporal graph ode networks for traffic flow forecasting. InProceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining. 364–373

    2021

  11. [11]

    Albert Gu and Tri Dao. 2023. Mamba: Linear-time sequence modeling with selective state spaces.arXiv preprint arXiv:2312.00752(2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  12. [12]

    Shengnan Guo, Youfang Lin, Ning Feng, Chao Song, and Huaiyu Wan. 2019. Attention based spatial-temporal graph convolutional networks for traffic flow forecasting. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 33. 922–929

    2019

  13. [13]

    Liangzhe Han, Bowen Du, Leilei Sun, Yanjie Fu, Yisheng Lv, and Hui Xiong. 2021. Dynamic and multi-faceted spatio-temporal deep learning for traffic speed forecasting. InProceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining. 547–555. Manuscript submitted to ACM From XXLTraffic to EvoXXLTraffic: Scaling Traffic Forecasting to Se...

    2021

  14. [14]

    Yuxin Jia, Youfang Lin, Xinyan Hao, Yan Lin, Shengnan Guo, and Huaiyu Wan. 2024. Witran: Water-wave information transmission and recurrent acceleration network for long-range time series forecasting.Advances in Neural Information Processing Systems36 (2024)

    2024

  15. [15]

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

    2023

  16. [16]

    Guangyin Jin, Yuxuan Liang, Yuchen Fang, Zezhi Shao, Jincai Huang, Junbo Zhang, and Yu Zheng. 2023. Spatio-temporal graph neural networks for predictive learning in urban computing: A survey.IEEE Transactions on Knowledge and Data Engineering(2023)

    2023

  17. [17]

    Guokun Lai, Wei-Cheng Chang, Yiming Yang, and Hanxiao Liu. 2018. Modeling long-and short-term temporal patterns with deep neural networks. InThe 41st international ACM SIGIR conference on research & development in information retrieval. 95–104

    2018

  18. [18]

    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. InInternational conference on machine learning. PMLR, 11906–11917

    2022

  19. [19]

    Hyunwook Lee, Seungmin Jin, Hyeshin Chu, Hongkyu Lim, and Sungahn Ko. 2021. Learning to Remember Patterns: Pattern Matching Memory Networks for Traffic Forecasting. InInternational Conference on Learning Representations

    2021

  20. [20]

    Hao Li, Jie Shao, Kewen Liao, and Mingjian Tang. 2022. Do Simpler Statistical Methods Perform Better in Multivariate Long Sequence Time-Series Forecasting?. InProceedings of the 31st ACM International Conference on Information & Knowledge Management. 4168–4172

    2022

  21. [21]

    Shiyang Li, Xiaoyong Jin, Yao Xuan, Xiyou Zhou, Wenhu Chen, Yu-Xiang Wang, and Xifeng Yan. 2019. Enhancing the locality and breaking the memory bottleneck of transformer on time series forecasting.Advances in Neural Information Processing Systems32 (2019), 5243–5253

    2019

  22. [22]

    Yanhong Li, Jack Xu, and David Anastasiu. 2024. Learning from Polar Representation: An Extreme-Adaptive Model for Long-Term Time Series Forecasting.Proceedings of the AAAI Conference on Artificial Intelligence38 (Mar. 2024), 171–179. doi:10.1609/aaai.v38i1.27768

    work page doi:10.1609/aaai.v38i1.27768 2024

  23. [23]

    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

  24. [24]

    Yang Lin, Irena Koprinska, and Mashud Rana. 2021. SSDNet: State space decomposition neural network for time series forecasting. In2021 IEEE International Conference on Data Mining (ICDM). IEEE, 370–378

    2021

  25. [25]

    Aoyu Liu and Yaying Zhang. [n. d.]. A General Spatio-Temporal Backbone with Scalable Contextual Pattern Bank for Urban Continual Forecasting. InThe Fourteenth International Conference on Learning Representations

  26. [26]

    Dachuan Liu, Jin Wang, Shuo Shang, and Peng Han. 2022. Msdr: Multi-step dependency relation networks for spatial temporal forecasting. In Proceedings of the 28th ACM SIGKDD conference on knowledge discovery and data mining. 1042–1050

    2022

  27. [27]

    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. InProceedings of the 32nd ACM international conference on information and knowledge management. 4125–4129

    2023

  28. [28]

    Shizhan Liu, Hang Yu, Cong Liao, Jianguo Li, Weiyao Lin, Alex X Liu, and Schahram Dustdar. 2021. Pyraformer: Low-complexity pyramidal attention for long-range time series modeling and forecasting. InInternational conference on learning representations

    2021

  29. [29]

    Xu Liu, Yutong Xia, Yuxuan Liang, Junfeng Hu, Yiwei Wang, Lei Bai, Chao Huang, Zhenguang Liu, Bryan Hooi, and Roger Zimmermann. 2024. Largest: A benchmark dataset for large-scale traffic forecasting.Advances in Neural Information Processing Systems36 (2024)

    2024

  30. [30]

    Yong Liu, Tengge Hu, Haoran Zhang, Haixu Wu, Shiyu Wang, Lintao Ma, and Mingsheng Long. 2024. iTransformer: Inverted Transformers Are Effective for Time Series Forecasting. InThe Twelfth International Conference on Learning Representations. https://openreview.net/forum?id= JePfAI8fah

    2024

  31. [31]

    Minbo Ma, Kai Tang, Huan Li, Fei Teng, Dalin Zhang, and Tianrui Li. 2025. Beyond fixed variables: Expanding-variate time series forecasting via flat scheme and spatio-temporal focal learning. InProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V. 2. 2054–2065

    2025

  32. [32]

    Yuqi Nie, Nam H Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. [n. d.]. A Time Series is Worth 64 Words: Long-term Forecasting with Transformers. InThe Eleventh International Conference on Learning Representations

  33. [33]

    Arian Prabowo, Hao Xue, Wei Shao, Piotr Koniusz, and Flora D Salim. 2024. Traffic forecasting on new roads using spatial contrastive pre-training (SCPT).Data Mining and Knowledge Discovery38 (2024), 913–937

    2024

  34. [34]

    Chao Shang, Jie Chen, and Jinbo Bi. 2021. Discrete Graph Structure Learning for Forecasting Multiple Time Series. InInternational Conference on Learning Representations

    2021

  35. [35]

    Zezhi Shao, Fei Wang, Yongjun Xu, Wei Wei, Chengqing Yu, Zhao Zhang, Di Yao, Guangyin Jin, Xin Cao, Gao Cong, et al. 2023. Exploring progress in multivariate time series forecasting: Comprehensive benchmarking and heterogeneity analysis.arXiv preprint arXiv:2310.06119(2023)

    work page arXiv 2023

  36. [36]

    Zezhi Shao, Fei Wang, Zhao Zhang, Yuchen Fang, Guangyin Jin, and Yongjun Xu. 2023. HUTFormer: Hierarchical U-Net Transformer for Long-Term Traffic Forecasting.arXiv preprint arXiv:2307.14596(2023)

    work page arXiv 2023

  37. [37]

    Zezhi Shao, Zhao Zhang, Wei Wei, Fei Wang, Yongjun Xu, Xin Cao, and Christian S Jensen. 2022. Decoupled dynamic spatial-temporal graph neural network for traffic forecasting.Proceedings of the VLDB Endowment15 (2022), 2733–2746

    2022

  38. [38]

    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. InProceedings of the AAAI conference on artificial intelligence, Vol. 34. 914–921

    2020

  39. [39]

    Binwu Wang, Yudong Zhang, Jiahao Shi, Pengkun Wang, Xu Wang, Lei Bai, and Yang Wang. 2023. Knowledge expansion and consolidation for continual traffic prediction with expanding graphs.IEEE Transactions on Intelligent Transportation Systems24, 7 (2023), 7190–7201. Manuscript submitted to ACM 26 Du Yin et al

    2023

  40. [40]

    Binwu Wang, Yudong Zhang, Xu Wang, Pengkun Wang, Zhengyang Zhou, Lei Bai, and Yang Wang. 2023. Pattern expansion and consolidation on evolving graphs for continual traffic prediction. InProceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 2223–2232

    2023

  41. [41]

    Huiqiang Wang, Jian Peng, Feihu Huang, Jince Wang, Junhui Chen, and Yifei Xiao. 2023. Micn: Multi-scale local and global context modeling for long-term series forecasting. InThe eleventh international conference on learning representations

    2023

  42. [42]

    Zhang, and JUN ZHOU

    Shiyu Wang, Haixu Wu, Xiaoming Shi, Tengge Hu, Huakun Luo, Lintao Ma, James Y. Zhang, and JUN ZHOU. 2024. TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting. InThe Twelfth International Conference on Learning Representations. https://openreview.net/forum? id=7oLshfEIC2

    2024

  43. [43]

    Gerald Woo, Chenghao Liu, Doyen Sahoo, Akshat Kumar, and Steven Hoi. 2023. Learning Deep Time-index Models for Time Series Forecasting. In Proceedings of the 40th International Conference on Machine Learning (Proceedings of Machine Learning Research, Vol. 202), Andreas Krause, Emma Brunskill, Kyunghyun Cho, Barbara Engelhardt, Sivan Sabato, and Jonathan S...

    2023

  44. [44]

    Haixu Wu, Tengge Hu, Yong Liu, Hang Zhou, Jianmin Wang, and Mingsheng Long. 2022. Timesnet: Temporal 2d-variation modeling for general time series analysis. InThe eleventh international conference on learning representations

    2022

  45. [45]

    Haixu Wu, Jiehui Xu, Jianmin Wang, and Mingsheng Long. 2021. Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting.Advances in Neural Information Processing Systems34 (2021), 22419–22430

    2021

  46. [46]

    Sifan Wu, Xi Xiao, Qianggang Ding, Peilin Zhao, Ying Wei, and Junzhou Huang. 2020. Adversarial sparse transformer for time series forecasting. Advances in neural information processing systems33 (2020), 17105–17115

    2020

  47. [47]

    Zonghan Wu, Shirui Pan, Guodong Long, Jing Jiang, Xiaojun Chang, and Chengqi Zhang. 2020. Connecting the dots: Multivariate time series forecasting with graph neural networks. InProceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining. 753–763

    2020

  48. [48]

    Zonghan Wu, Shirui Pan, Guodong Long, Jing Jiang, and Chengqi Zhang. 2019. Graph wavenet for deep spatial-temporal graph modeling. In Proceedings of the 28th International Joint Conference on Artificial Intelligence. 1907–1913

    2019

  49. [49]

    Du Yin, Hao Xue, Arian Prabowo, Shuang Ao, and Flora Salim. 2025. XXLTraffic: Expanding and Extremely Long Traffic forecasting beyond test adaptation. InProceedings of the 33rd ACM International Conference on Advances in Geographic Information Systems. 511–521

    2025

  50. [50]

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

    2018

  51. [51]

    Chengqing Yu, Fei Wang, Zezhi Shao, Tao Sun, Lin Wu, and Yongjun Xu. 2023. Dsformer: A double sampling transformer for multivariate time series long-term prediction. InProceedings of the 32nd ACM International Conference on Information and Knowledge Management. 3062–3072

    2023

  52. [52]

    Ailing Zeng, Muxi Chen, Lei Zhang, and Qiang Xu. 2023. Are transformers effective for time series forecasting?. InProceedings of the AAAI conference on artificial intelligence, Vol. 37. 11121–11128

    2023

  53. [53]

    Haoyu Zhang, Hao Miao, Xinke Jiang, Yuchen Fang, and Yifan Zhang. 2025. Strap: Spatio-temporal pattern retrieval for out-of-distribution generalization.Advances in Neural Information Processing Systems38 (2025), 118006–118041

    2025

  54. [54]

    Ling Zhao, Yujiao Song, Chao Zhang, Yu Liu, Pu Wang, Tao Lin, Min Deng, and Haifeng Li. 2019. T-GCN: A temporal graph convolutional network for traffic prediction.IEEE transactions on intelligent transportation systems21, 9 (2019), 3848–3858

    2019

  55. [55]

    Yusheng Zhao, Xiao Luo, Wei Ju, Chong Chen, Xian-Sheng Hua, and Ming Zhang. 2023. Dynamic hypergraph structure learning for traffic flow forecasting. In2023 IEEE 39th International Conference on Data Engineering (ICDE). IEEE, 2303–2316

    2023

  56. [56]

    Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang. 2021. Informer: Beyond efficient transformer for long sequence time-series forecasting. InProceedings of the AAAI conference on artificial intelligence, Vol. 35. 11106–11115

    2021

  57. [57]

    Tian Zhou, Ziqing Ma, Qingsong Wen, Xue Wang, Liang Sun, and Rong Jin. 2022. Fedformer: Frequency enhanced decomposed transformer for long-term series forecasting. InInternational Conference on Machine Learning. PMLR, 27268–27286

    2022

  58. [58]

    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. InProceedings of the 17th ACM International Conference on Web Search and Data Mining. 1032–1041. Received 1 June 2026; rev...

    2024