Pith. sign in

REVIEW 5 major objections 6 minor 37 references

FasterSTS: A Faster Spatio-Temporal Synchronous Graph Convolutional Networks for Traffic flow Forecasting

T0 review · 5 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read FasterSTS claims state-of-the-art traffic forecasts while cutting graph convolution cost from quadratic to linear.

desk verdict The efficiency claim is real and worth engaging, but the paper's headline prediction claim is contradicted by its own Table IV and the equations cannot be reconstructed. read the letter →

arxiv 2501.00756 v1 pith:EHO5YK27 submitted 2025-01-01 cs.LG

Ben-Ao Dai , Nengchao Lyu , Yongchao Miao This is my paper · ORCID
classification cs.LG
keywords trafficflowforecastingspatio-temporalsynchronousmodelinggraphconvolutionalnetworksfastcomputationadaptivelow-ranknodeaggregationPEMSdatasets
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

FasterSTS is a traffic-flow forecasting model whose central claim is that spatio-temporal correlations can be modeled synchronously, inside a single graph-convolution operation, rather than by separate spatial and temporal blocks. The paper introduces a fast graph computation that replaces the dense $N \times N$ adjacency operation with an $N \times n$ aggregation step followed by an $n \times N$ projection, cutting the graph-convolution cost from $O(N^2)$ to $O(nN)$. On four public California highway datasets (PEMS03, PEMS04, PEMS07, PEMS08), the paper claims lower MAE, MAPE, and RMSE than ten baseline models, while using no multi-head self-attention, no recurrent neural network, and no temporal convolutional network. If those comparisons hold under a common evaluation protocol, accurate spatio-temporal forecasting would become practical on resource-limited devices and inside real-time loops such as reinforcement-learning controllers.

What carries the argument

The load-bearing object is the fast graph computation, which splits the conventional graph operation into node information aggregation, a trainable matrix $E \in \mathbb{R}^{N \times n}$ with $n \ll N$, applied as a softmax-weighted Hadamard product, and node representation projection, a $1\times1$ convolution that lifts the $n$-dimensional aggregated representation back to $N$ dimensions. This reduces graph computation from $O(N^2)$ to $O(nN)$. The second central object is the spatio-temporal synchronous graph convolution kernel $\Psi$, built from a static learnable embedding and a dynamic input-derived embedding, which lets graph convolution itself carry temporal correlation instead of delegating it to RNNs, TCNs, or attention. The adaptive graph is assembled from global and local node embeddings, one per hidden dimension, so the dynamic kernel can gate and adjust spatial features during convolution.

What would settle it

Rerun the ten baselines on the same machine, with the same data splits, normalization, and evaluation script used for FasterSTS, and recompute Tables III and IV; if any baseline beats FasterSTS's reported average MAE, MAPE, or RMSE on any dataset, the 'leads all baselines' claim is refuted. A cheaper check is already available in the paper: in Table IV, the STWave row has lower average MAE and MAPE than FasterSTS on PEMS07.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that the expensive parts of a spatio-temporal synchronous graph convolutional network can be replaced without giving up accuracy: the graph aggregation is factored into two low-rank steps, and the convolution kernel is made time-aware by splitting it into a static adaptive embedding and a dynamic adaptive embedding whose combination forms a spatio-temporal synchronous kernel $\Psi$. The static part models stable temporal dependencies and feature mapping; the dynamic part, derived from the input through a $1\times1$ convolution and a softmax, adjusts the kernel and the adaptive graph over time. The authors argue that this lets one graph-convolution layer capture static and dynamic spatial correlation, static and dynamic temporal correlation, and the heterogeneity among them. They claim that this design outperforms ten prior models in three error metrics on four real-world traffic datasets while reducing training time and GPU memory several-fold relative to the spatio-temporal synchronous baselines.

Load-bearing premise

The load-bearing premise is that every baseline number in Tables III and IV was produced under exactly the same protocol as FasterSTS, the same 6:2:2 data split, the same 12-step history, the same normalization, and the same evaluation code, because the paper takes those baseline numbers from earlier papers rather than rerunning the models in its own environment.

Editorial extensions

If this is right

  • If the reported numbers survive a shared evaluation protocol, FasterSTS would be a spatio-temporal synchronous model that matches attention-based and wavelet-based transformers on the PEMS datasets without using attention at all.
  • The reduction from $O(N^2)$ to $O(nN)$ would let the same architecture scale to city-scale networks with tens of thousands of sensors on a single GPU.
  • Because the model uses only embeddings, convolutions, and residual normalizations, its forward pass is simpler to compile for edge devices and fast enough to sit inside reinforcement-learning loops.
  • The ablation study indicates that the fast graph operation is mostly a speed device: removing it leaves MAE essentially unchanged, so the accuracy gains are attributed to the synchronous kernel and the adaptive graphs rather than to the low-rank factorization.

Reading between the lines

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

  • Editorial inference: the low-rank aggregation pattern (project to $n$ dimensions, aggregate, project back) is not specific to traffic data, so it could be dropped into other learned-adjacency graph convolutions where $N$ is large, such as recommender systems or molecular graphs.
  • Editorial observation: the paper's own Table IV shows STWave with lower average MAE (19.60 vs 19.67) and MAPE (8.21 vs 8.30) on PEMS07, so the sentence claiming that FasterSTS leads all baselines in all three metrics on all four datasets is not supported by the reported numbers; the supportable claim is that FasterSTS is competitive with STWave and better in RMSE.
  • Editorial inference: the static-plus-dynamic kernel factorization suggests a separable design principle, use a low-cost static kernel to anchor stable patterns and a data-dependent kernel to modulate them, which could be tested on other spatio-temporal tasks such as energy load or crowd-flow forecasting.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The paper proposes FasterSTS, a spatio-temporal synchronous graph convolutional network for traffic flow forecasting. The core ideas are (1) a 'fast graph computation' that replaces the N×N adaptive adjacency with an N×n low-rank aggregation followed by a 1×1 convolution projection, reducing graph convolution cost from O(N²) to O(nN); (2) a spatio-temporal synchronous convolution kernel Ψ built from static learnable embeddings and a data-dependent dynamic part, intended to capture temporal correlations during graph convolution; and (3) per-hidden-dimension adaptive graphs using global and local node embeddings. The model is evaluated on PEMS03/04/07/08 and compared with eight baselines on MAE, MAPE, and RMSE, plus an efficiency comparison in Table VI. The paper claims FasterSTS 'significantly leads all baseline models in three evaluation indexes on all four datasets' (Section IV-D).

Significance. If substantiated, the model would be a meaningful practical contribution: it promises state-of-the-art forecasting accuracy at substantially lower training and inference cost, which is relevant for real-time and resource-constrained applications. The low-rank graph aggregation idea is intuitively appealing and the efficiency numbers in Table VI (e.g., 37 s/epoch and 1759 MB GPU memory versus 256 s and 6789 MB for STGODE on PEMS07) are suggestive. However, the significance is heavily undercut by the fact that the paper's own Tables III–IV contradict the headline empirical claim, and by the under-specified formal definition of the architecture. The efficiency advantage alone, without a reliable accuracy claim, is not sufficient to establish the paper's central thesis.

major comments (5)
  1. [IV-D, Table IV] The central claim that FasterSTS 'significantly leads all baseline models in three evaluation indexes on all four datasets' is directly contradicted by the paper's own Table IV for PEMS07. On the average row, STWave achieves MAE 19.60 and MAPE 8.21%, while FasterSTS achieves MAE 19.67 and MAPE 8.30%; STWave is also better on MAE and MAPE at the 15-, 30-, and 60-minute horizons. Only RMSE favors FasterSTS. Thus the stated claim is false as written. If the table entries are accurate, the conclusion must be narrowed; if they are not accurate, the paper has not supported the claim with corrected data.
  2. [III-E, Eqs. (6)–(9)] The formal specification of the spatio-temporal synchronous convolution kernel is internally inconsistent and cannot be reconstructed from the text. The kernel Ψ is first defined as Ψ ∈ R^{T din × T dout}, then later as Ψ ∈ R^{T din × dΨ}, with dΨ said to be much less than T dout; Eq. (6) writes X ∈ R^{N × dΨ} immediately after stating X ∈ R^{N × T × din}. The fast graph computation in Eq. (3) defines A ∈ R^{N × n} via A(i,j)=[Softmax(E_{:,j})]_i, but Eq. (13) applies A ⊙ X with X of shape N × T × C, which is a dimension mismatch unless additional reshaping is assumed and never stated. The 'dimension transformation' R() is described only verbally, and the sizes of CΨ and cΨ are not given. These omissions mean the proposed architecture is not precisely defined.
  3. [IV-B and IV-C, Tables III–IV] The comparison against baselines is not established under a shared protocol. Section IV-B states the 6:2:2 split for FasterSTS, but the paper does not state that the baseline metrics in Tables III–IV were obtained by rerunning those models under the identical preprocessing, input length (12 steps), normalization, evaluation horizon, or random seed protocol. The baseline numbers appear to be imported from prior publications (e.g., the STWave values match its ICDE paper). Without shared-protocol reruns or a statement that all baselines use the same pipeline, the claim of 'significantly leads all baselines' is unsupported even setting aside the PEMS07 counterexample.
  4. [IV-H] The complexity analysis is not rigorous. The paper claims the spatio-temporal kernel cost is reduced 'from T din × T dout' to 'T dindΨ + T doutdΨ', but the parenthetical 'T din × T dout is equal to T din T 2 dout + T dout T 2 din' is syntactically meaningless, and no derivation is given for the claimed costs. The analysis also omits the costs of the dynamic kernel generation steps (Eqs. (7)–(9)), the Softmax, the gating operation, and the per-hidden-dimension adaptive graph construction. The O(N²)-to-O(nN) reduction for graph aggregation is plausible, but the overall claim that the model is 'reduced from quadratic to linear level' is not demonstrated.
  5. [IV-E, Table V] The ablation results do not support the statement that the fast graph operation 'achieves a similar effect to traditional graph operations'. In Table V, the variant without fast graph computation (w/o FGC) is better than FasterSTS on all three metrics on PEMS04 (MAE 18.40 vs. 18.49, RMSE 29.90 vs. 29.92, MAPE 12.16 vs. 12.21). The paper's claim of parity is contradicted by its own numbers, and this weakens the argument that the efficiency gain is obtained without sacrificing accuracy.
minor comments (6)
  1. [Algorithm 1, line 15] The algorithm says 'the parameters are updated using gradient descent to minimize the Eq.11', but Eq. (11) is a residual connection, not a loss function; the loss is defined in Eq. (17). The reference should be to Eq. (17).
  2. [III-E, 'Reinforcement Learning Component'] The 'Reinforcement Learning Component' is described as two fully connected layers with ReLU; no reinforcement learning signal, policy, or reward is involved. The name is misleading and should be changed to something like 'temporal embedding projection'.
  3. [II-C] Two consecutive paragraphs in Section II-C are duplicated verbatim, beginning 'Compared with the spatio-temporal asynchronous modeling methods mentioned above...'. One copy should be removed.
  4. [References] References [35] and [39] are identical (both cite Chen et al., 'Bidirectional spatial-temporal adaptive transformer'); [35] is used for Bi-STAT in the tables and [39] is also cited for Bi-STAT. Duplicate references should be consolidated.
  5. [Throughout] There are numerous typos and inconsistent notations: 'Sof tmax' for Softmax, 'caputred' for captured, 'Traning' and 'Interferce' in Table VI, 'a 1×1 convolutional layer (whose time complexity is n ∗ N)' which conflates the convolution kernel with the complexity, and the undefined symbol K in the abstract's complexity claim (O(KN)).
  6. [IV-F] The hyperparameter analysis refers to Fig. 5, but no figure content is described and no quantitative results are given; the text only states generic overfitting behavior. Please provide actual plots or tables with the sensitivity values.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation; the architecture is specified and empirically evaluated, though one self-citation is used for design inspiration and baseline comparisons are imported without a shared protocol.

full rationale

FasterSTS derives no numerical constant from the data it predicts; all learned parameters are trained with the MAE loss (Eq. 17) and the reported metrics (Eqs. 18-20) are evaluated on held-out test splits. The spatio-temporal synchronous kernel is defined by explicit equations (Eqs. 6-9), and the fast graph computation is defined by the aggregation and projection operations of Eqs. 3-4 and 13-14. The sentence 'Our inspiration comes from HTVGNN[12]' is a self-citation to prior work by an overlapping author, but it is not invoked as a proof, uniqueness theorem, or externally imposed constraint; the present paper states its own construction and tests it directly. Similarly, the citation [12] for adaptive graphs assigns different embedding dimensions per hidden dimension, but the mechanism is implemented and ablated here. Importing baseline numbers from earlier papers without rerunning baselines under the identical 6:2:2 protocol may weaken the empirical comparison, but this is a protocol and correctness concern, not a circular derivation. The paper's own Table IV appears to contradict the Section IV-D claim that FasterSTS 'significantly leads all baseline models in three evaluation indexes on all four datasets' (STWave has lower MAE and MAPE on PEMS07), but an internally inconsistent claim is not the same as a claim that reduces to its inputs by construction.

Assumptions & free parameters 5 free parameters · 4 assumptions · 3 invented entities

The model's contribution is architectural; it introduces several learned components and hyperparameters. The main unsupported leaps are the low-rank approximation validity and the kernel factorization, both justified only by the paper's own experiments.

free parameters (5)
  • n (bottleneck rank of fast graph computation) = not reported
    Controls the tradeoff between complexity and approximation quality; the paper never specifies n, only says n << N.
  • D (adaptive graph embedding dimension) = 6, 8, or 10 depending on dataset (Table II)
    Chosen manually per dataset; larger D leads to overfitting, per the paper's own hyperparameter analysis.
  • L (number of STSGCL layers) = 2, 4, 6 per dataset (Table II)
    Selected by hand; too many layers increase complexity and overfit.
  • H (hidden dimension) = 32 for all datasets
    Set in Table II; fixed across datasets.
  • learning rate = 0.001
    Standard choice; not tuned.
assumptions (4)
  • standard math First-order Chebyshev approximation of graph convolution (Eq. 5) is a valid spectral filter.
    The paper adopts the 1st ChebNet approximation from Kipf and Welling without proof; standard GCN background.
  • domain assumption The low-rank N-by-n factor E followed by 1x1 convolution can approximate the full N-by-N graph transform.
    No theoretical bound or experiment measuring approximation error is given; the ablation only shows equivalent accuracy on PEMS04.
  • ad hoc to paper The spatio-temporal synchronous kernel Psi can be factorized into a static embedding part and a dynamic data-dependent part.
    This factorization is introduced specifically for this model; no independent grounding.
  • domain assumption Softmax-normalized trainable embeddings define a valid adjacency structure.
    Standard in Graph WaveNet and related work; assumed to capture hidden spatial dependencies.
invented entities (3)
  • Spatio-temporal synchronous graph convolution kernel Psi in R^{T*din x T*dout}
    purpose: Encodes temporal correlations into the graph convolution kernel so that spatial and temporal features are processed simultaneously.
    A new kernel shape proposed here; only evaluated within this paper's experiments.
  • RL Component
    purpose: Project the temporal embedding Upsilon to the kernel dimension using two fully connected layers; named 'reinforcement learning' but no RL mechanism is used.
    The name is misleading; the component is a standard MLP and no reward signal or policy is defined.
  • Per-hidden-dimension adaptive graphs from global plus local embeddings
    purpose: Model static and dynamic spatial correlations separately for each hidden dimension.
    Extension of Graph WaveNet's adaptive graph; the paper adds per-dimension local embeddings.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FasterSTS: A Faster Spatio-Temporal Synchronous Graph Convolutional Networks for Traffic flow Forecasting." pith.science (2026). https://pith.science/paper/EHO5YK27

@misc{pith2026250100756,
  author       = {Pith},
  title        = {Pith review of: FasterSTS: A Faster Spatio-Temporal Synchronous Graph Convolutional Networks for Traffic flow Forecasting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EHO5YK27}},
  note         = {Machine review of arXiv:2501.00756}
}
read the original abstract

Accurate traffic flow prediction heavily relies on the spatio-temporal correlation of traffic flow data. Most current studies separately capture correlations in spatial and temporal dimensions, making it difficult to capture complex spatio-temporal heterogeneity, and often at the expense of increasing model complexity to improve prediction accuracy. Although there have been groundbreaking attempts in the field of spatio-temporal synchronous modeling, significant limitations remain in terms of performance and complexity control.This study proposes a quicker and more effective spatio-temporal synchronous traffic flow forecast model to address these issues.

Figures

Figures reproduced from arXiv: 2501.00756 by the authors.

Figure 1
Figure 1. The process of graph computation with different methods [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. The architecture of FasterSTS E. Spatio-Temporal synchronous Graph Convolutional Layers As shown in the Fig.2, FasterSTS is composed of multi￾ple spatio-temporal synchronous graph convolutional layers. The spatio-temporal synchronous graph convolution is the key component of FasterSTS, denoted as ST G(). To model the temporal correlation in the graph convolution process, ST G() reforms the traditional graph convolut… view at source ↗
Figure 3
Figure 3. The computation process of the spatio-temporal syn [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 29 canonical work pages

  1. [1]

    Spatio-temporal adaptive embedding makes vanilla transformer sota for traffic forecasting,

    H. Liu, Z. Dong, R. Jiang, J. Deng, J. Deng, Q. Chen, and X. Song, “Spatio-temporal adaptive embedding makes vanilla transformer sota for traffic forecasting,” in Pro- ceedings of the 32nd ACM international conference on information and knowledge management , 2023, pp. 4125–4129

    work page 2023

  2. [2]

    A comprehensive survey on graph neural networks,

    Z. Wu, S. Pan, F. Chen, G. Long, C. Zhang, and S. Y . Philip, “A comprehensive survey on graph neural networks,” IEEE transactions on neural networks and learning systems, vol. 32, no. 1, pp. 4–24, 2020

    work page 2020

  3. [3]

    Stfgcn: Spatial–temporal fusion graph convolu- tional network for traffic prediction,

    H. Li, J. Liu, S. Han, J. Zhou, T. Zhang, and C. P. Chen, “Stfgcn: Spatial–temporal fusion graph convolu- tional network for traffic prediction,” Expert Systems with Applications, vol. 255, p. 124648, 2024. 12

    work page 2024

  4. [4]

    Short-term traffic forecasting: Where we are and where we’re going,

    E. I. Vlahogianni, M. G. Karlaftis, and J. C. Golias, “Short-term traffic forecasting: Where we are and where we’re going,” Transportation Research Part C: Emerging Technologies, vol. 43, pp. 3–19, 2014

    work page 2014

  5. [5]

    Spatiotemporal gaussian process kalman filter for mo- bile traffic prediction,

    Y . Cai, P. Cheng, M. Ding, Y . Chen, Y . Li, and B. Vucetic, “Spatiotemporal gaussian process kalman filter for mo- bile traffic prediction,” in 2020 IEEE 31st Annual In- ternational Symposium on Personal, Indoor and Mobile Radio Communications. IEEE, 2020, pp. 1–6

    work page 2020

  6. [6]

    Short-term traffic and travel time prediction models,

    J. Van Lint and C. Van Hinsbergen, “Short-term traffic and travel time prediction models,” Artificial Intelligence Applications to Critical Transportation Issues , vol. 22, no. 1, pp. 22–41, 2012

    work page 2012

  7. [7]

    Supervised weighting-online learning algorithm for short-term traffic flow prediction,

    Y .-S. Jeong, Y .-J. Byon, M. M. Castro-Neto, and S. M. Easa, “Supervised weighting-online learning algorithm for short-term traffic flow prediction,” IEEE Transactions on Intelligent Transportation Systems, vol. 14, no. 4, pp. 1700–1707, 2013

    work page 2013

  8. [8]

    Long short-term memory,

    N. Computation, “Long short-term memory,” Neural Comput, vol. 9, pp. 1735–1780, 2016

    work page 2016

Show all 37 references
  1. [9]

    Short-term traffic flow prediction at isolated intersections based on parallel multi-task learning,

    B.-L. Ye, S. Zhu, L. Li, and W. Wu, “Short-term traffic flow prediction at isolated intersections based on parallel multi-task learning,” Systems Science & Control Engi- neering, vol. 12, no. 1, p. 2316160, 2024

    2024

  2. [10]

    An attention-based deep learning model for traffic flow prediction using spatiotemporal features towards sustainable smart city,

    B. Vijayalakshmi, K. Ramar, N. Jhanjhi, S. Verma, M. Kaliappan, K. Vijayalakshmi, S. Vimal, and U. Ghosh, “An attention-based deep learning model for traffic flow prediction using spatiotemporal features towards sustainable smart city,” International Journal of Communication S...

    2021

  3. [11]

    Pre-training enhanced spatial-temporal graph neural network for mul- tivariate time series forecasting,

    Z. Shao, Z. Zhang, F. Wang, and Y . Xu, “Pre-training enhanced spatial-temporal graph neural network for mul- tivariate time series forecasting,” in Proceedings of the 28th ACM SIGKDD conference on knowledge discovery and data mining , 2022, pp. 1567–1577

    2022

  4. [12]

    A novel hybrid time-varying graph neural network for traffic flow forecasting,

    B. A. Dai and B.-L. Ye, “A novel hybrid time-varying graph neural network for traffic flow forecasting,” arXiv preprint arXiv:2401.10155, 2024

    2024 arXiv

  5. [14]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural informa- tion processing systems , vol. 30, 2017

    2017

  6. [16]

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

    B. Yu, H. Yin, and Z. Zhu, “Spatio-temporal graph convolutional networks: A deep learning framework for traffic forecasting,” arXiv preprint arXiv:1709.04875 , 2017

    2017 arXiv

  7. [17]

    Diffusion con- volutional recurrent neural network: Data-driven traffic forecasting,

    Y . Li, R. Yu, C. Shahabi, and Y . Liu, “Diffusion con- volutional recurrent neural network: Data-driven traffic forecasting,” arXiv preprint arXiv:1707.01926 , 2017

    2017 arXiv

  8. [18]

    Graph wavenet for deep spatial-temporal graph mod- eling,

    Z. Wu, S. Pan, G. Long, J. Jiang, and C. Zhang, “Graph wavenet for deep spatial-temporal graph mod- eling,” arXiv preprint arXiv:1906.00121 , 2019

    1906 arXiv

  9. [19]

    Traffic flow prediction based on multi-spatial-temporal graph- convolutional network,

    D. Junming, C. Yang, S. Qinqin, and S. Quan, “Traffic flow prediction based on multi-spatial-temporal graph- convolutional network,” Application Research of Com- pute, vol. 39, no. 3, p. 5, 2022

    2022

  10. [20]

    Comparative analysis of time series and machine learn- ing models for air quality prediction utilizing iot data,

    G. V onitsanos, T. Panagiotakopoulos, and A. Kameas, “Comparative analysis of time series and machine learn- ing models for air quality prediction utilizing iot data,” in IFIP International Conference on Artificial Intelligence Applications and Innovations. Springer, 2024, pp....

    2024

  11. [21]

    The impact of cutting-edge technologies on smart city supply chain: A systematic literature review of the evidence and implications,

    B. Najafi, A. Najafi, F. Madanchi, H. Maghroor, and H. Taherdoost, “The impact of cutting-edge technologies on smart city supply chain: A systematic literature review of the evidence and implications,” IEEE Engineering Management Review, 2024

    2024

  12. [22]

    Stghtn: Spatial-temporal gated hybrid transformer network for traffic flow forecasting,

    J. Liu, Y . Kang, H. Li, H. Wang, and X. Yang, “Stghtn: Spatial-temporal gated hybrid transformer network for traffic flow forecasting,” Applied Intelligence , vol. 53, no. 10, pp. 12 472–12 488, 2023

    2023

  13. [23]

    Spatial- temporal synchronous graph convolutional networks: A new framework for spatial-temporal network data fore- casting,

    C. Song, Y . Lin, S. Guo, and H. Wan, “Spatial- temporal synchronous graph convolutional networks: A new framework for spatial-temporal network data fore- casting,” in Proceedings of the AAAI conference on artificial intelligence, vol. 34, no. 01, 2020, pp. 914–921

    2020

  14. [24]

    Spatial-temporal fusion graph neural networks for traffic flow forecasting,

    M. Li and Z. Zhu, “Spatial-temporal fusion graph neural networks for traffic flow forecasting,” in Proceedings of the AAAI conference on artificial intelligence , vol. 35, no. 5, 2021, pp. 4189–4196

    2021

  15. [25]

    Learning con- volutional neural networks for graphs,

    M. Niepert, M. Ahmed, and K. Kutzkov, “Learning con- volutional neural networks for graphs,” in International conference on machine learning . PMLR, 2016, pp. 2014–2023

    2016

  16. [27]

    Geometric deep learning on graphs and manifolds using mixture model cnns,

    F. Monti, D. Boscaini, J. Masci, E. Rodola, J. Svoboda, and M. M. Bronstein, “Geometric deep learning on graphs and manifolds using mixture model cnns,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2017, pp. 5115–5124

    2017

  17. [28]

    Spectral networks and locally connected networks on graphs,

    J. Bruna, W. Zaremba, A. Szlam, and Y . LeCun, “Spectral networks and locally connected networks on graphs,” arXiv preprint arXiv:1312.6203 , 2013

    2013 arXiv

  18. [29]

    Con- volutional neural networks on graphs with fast localized spectral filtering,

    M. Defferrard, X. Bresson, and P. Vandergheynst, “Con- volutional neural networks on graphs with fast localized spectral filtering,” Advances in neural information pro- cessing systems, vol. 29, 2016

    2016

  19. [30]

    Attention based spatial-temporal graph convolutional networks for traffic flow forecasting,

    S. Guo, Y . Lin, N. Feng, C. Song, and H. Wan, “Attention based spatial-temporal graph convolutional networks for traffic flow forecasting,” in Proceedings of the AAAI conference on artificial intelligence , vol. 33, no. 01, 2019, pp. 922–929

    2019

  20. [31]

    Learning dynamic and hier- archical traffic spatiotemporal features with transformer,

    H. Yan, X. Ma, and Z. Pu, “Learning dynamic and hier- archical traffic spatiotemporal features with transformer,” IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 11, pp. 22 386–22 399, 2021

    2021

  21. [32]

    Learning dynamics and heterogeneity of spatial-temporal graph 13 data for traffic forecasting,

    S. Guo, Y . Lin, H. Wan, X. Li, and G. Cong, “Learning dynamics and heterogeneity of spatial-temporal graph 13 data for traffic forecasting,” IEEE Transactions on Knowledge and Data Engineering , vol. 34, no. 11, pp. 5415–5428, 2021

    2021

  22. [33]

    Spatial- temporal graph ode networks for traffic flow forecasting,

    Z. Fang, Q. Long, G. Song, and K. Xie, “Spatial- temporal graph ode networks for traffic flow forecasting,” in Proceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining , 2021, pp. 364–373

    2021

  23. [34]

    Semi-supervised classifica- tion with graph convolutional networks,

    T. N. Kipf and M. Welling, “Semi-supervised classifica- tion with graph convolutional networks,” arXiv preprint arXiv:1609.02907, 2016

    2016 arXiv

  24. [35]

    Bidirectional spatial-temporal adaptive transformer for urban traffic flow forecasting,

    C. Chen, Y . Liu, L. Chen, and C. Zhang, “Bidirectional spatial-temporal adaptive transformer for urban traffic flow forecasting,” IEEE Transactions on Neural Net- works and Learning Systems , vol. 34, no. 10, pp. 6913– 6925, 2022

    2022

  25. [36]

    Stgformer: Efficient spatiotem- poral graph transformer for traffic forecasting,

    H. Wang, J. Chen, T. Pan, Z. Dong, L. Zhang, R. Jiang, and X. Song, “Stgformer: Efficient spatiotem- poral graph transformer for traffic forecasting,” arXiv preprint arXiv:2410.00385, 2024

    2024 arXiv

  26. [37]

    Vector autoregressive models for multivariate time series,

    E. Zivot and J. Wang, “Vector autoregressive models for multivariate time series,” Modeling financial time series with S-PLUS®, pp. 385–429, 2006

    2006

  27. [38]

    When spatio-temporal meet wavelets: Dis- entangled traffic forecasting via efficient spectral graph attention networks,

    Y . Fang, Y . Qin, H. Luo, F. Zhao, B. Xu, L. Zeng, and C. Wang, “When spatio-temporal meet wavelets: Dis- entangled traffic forecasting via efficient spectral graph attention networks,” in 2023 IEEE 39th International Conference on Data Engineering (ICDE) . IEEE, 2023, pp. 517–529

    2023

  28. [39]

    Bidirectional spatial-temporal adaptive transformer for urban traffic flow forecasting,

    C. Chen, Y . Liu, L. Chen, and C. Zhang, “Bidirectional spatial-temporal adaptive transformer for urban traffic flow forecasting,” IEEE Transactions on Neural Net- works and Learning Systems , vol. 34, no. 10, pp. 6913– 6925, 2022. Ben-Ao Dai received the M.S. degree in Elec- ...

    2022

  29. [2008]

    His research interests include advanced driver assistance system (ADAS) and intelligent vehicle (IV), traffic safety operation management, and traffic safety evaluation. He has hosted 4 National Nature Science Funds related to driving behavior and traffic safety; he has finish...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.