Pith. sign in

REVIEW 3 major objections 6 minor 43 references

F2STNet: Fair and Federated Spectral-Temporal Modeling for Graph Forecasting

T0 review · 3 major / 6 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 F2STNet claims that a single federated architecture—combining a truncated graph-Fourier basis, a diagonal state-space temporal encoder, and loss-aware client reweighting—achieves both better forecasting accuracy and better client-level…

desk verdict Coherent hybrid model with a sensible fairness aggregation, but the empirical evidence is thinner than the claims suggest—baseline variance and the FL backbone need to be reported. read the letter →

arxiv 2608.09082 v1 pith:5DZVS7KH submitted 2026-08-10 cs.LG

Jiayi Zhang , Jinfeng Xu , Hewei Wang , Siyuan Cen , Haidong Huang , Yiyao Zhan , Zheyu Chen , Jinjiang You
show 2 more authors
Ai Jian Edith C. H. Ngai
This is my paper · ORCID
classification cs.LG
keywords federatedlearningspatiotemporalforecastinggraphneuralnetworksFouriertransformstate-spacemodelsfairness-awareaggregationtrafficairquality
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

F2STNet is a single federated architecture that forecasts graph-structured time series by combining a truncated graph-Fourier basis, a diagonal state-space temporal encoder, and fairness-aware client aggregation. The paper's central claim is that this combination beats the compared baselines on three real-world benchmarks (best RMSE, MAE, and MAPE on PeMS04 and KnowAir; best MAE on HZMetro) and, under federated splits of PeMS04, is more equitable across clients in worst-case error and client-to-client dispersion. If true, it means that privacy-preserving, resource-constrained forecasting can get both accuracy and fairness from one compact model instead of having to choose between them.

What carries the argument

The central carrying object is the truncated graph-Fourier basis $U_F \in \mathbb{R}^{N\times F}$, built from the eigendecomposition of the normalized graph Laplacian with self-loops $\tilde{L} = I - \tilde{D}^{-1/2}\tilde{A}\tilde{D}^{-1/2} = U\Lambda U^\top$. Keeping $F$ eigenvectors and forming $S_{b,t,n,f} = (U_F)_{n,f} (U_F^\top X)_{b,t,f}$ gives each node a frequency-coordinate vector that a shared MLP embeds, letting the model see local and global graph-frequency structure without learning the projection. The other key piece is the diagonal state-space temporal layer: for each node, a per-step scan $h_t = \bar{A}_t \odot h_{t-1} + \bar{B}_t \odot z_t$ with $\bar{A}_t = \exp(\Delta_t \odot A)$ and $\bar{B}_t = (\bar{A}_t - 1) \oslash A \odot B_t$, outputting only the final state. This gives $O(T)$ temporal complexity, and the ablation shows it is the component whose removal hurts most. FFA's weighting $q_k^t = p_k^t(1 + \lambda_t r_k^t)$, with $\lambda_t$ increasing under a reverse-annealing schedule, is the fairness mechanism.

What would settle it

Run FFA on a fixed PeMS04 split but replace each client's validation loss with a noisy surrogate matched in mean and variance to the true loss. If worst-client MAE and client-wise standard deviation stop improving over FedAvg, then FFA's reported fairness gains depend on validation loss reliability, a sensitivity the paper itself acknowledges.

Watch

Extended reading notes

Core claim

F2STNet claims that spectral and state-space temporal modeling plus loss-based client reweighting deliver both accuracy and fairness in federated graph forecasting. The spectral branch projects each node's signal onto a truncated graph-Fourier basis $U_F$ precomputed from the shared normalized graph Laplacian, giving every node a frequency profile that a shared MLP embeds; the temporal branch runs a diagonal state-space layer with input-dependent parameters over each node's sequence, using only the final latent state; a graph convolution then mixes spatial neighborhoods, and a linear head produces the $H$-step forecast. In the federated setting, FFA replaces FedAvg's sample-count weights with weights adjusted by relative validation-loss deviation, scaled by an annealing coefficient that grows over rounds. The paper reports that F2STNet achieves the best RMSE, MAE, or MAPE on PeMS04 and KnowAir, the best MAE on HZMetro, and the best composite utility-fairness score on federated PeMS04, with lower worst-client RMSE and lower client-wise standard deviation than all compared aggregation methods.

Load-bearing premise

The load-bearing premise is that every federated client observes the same set of $N$ nodes with the same adjacency matrix, because the graph-Fourier basis and the graph convolution are global and precomputed once; if clients hold different sensors or different road networks, the spectral projection and convolution cannot be applied as written.

Editorial extensions

If this is right

  • If the accuracy claims hold, federated traffic and environmental forecasting can drop heavy recurrent or attention modules: the diagonal state-space layer already covers long-range dependence at linear cost.
  • FFA-style reweighting could serve as a drop-in replacement for FedAvg in other federated regression tasks, improving worst-client performance without adding protected attributes or a separate fairness objective.
  • Because the graph-Fourier basis is precomputed once from shared topology, the spectral part adds no per-round communication or runtime overhead, so the efficiency gain transfers directly to bandwidth-constrained deployments.
  • The reported gains across traffic, metro, and air-quality data suggest the architecture is not tied to one sensor modality, though the federated analysis centres on PeMS04.
  • The composite utility-fairness score $\Delta$ defined in the paper offers a single number that captures both average accuracy and client disparity, which could become a standard reporting metric for federated forecasting.

Reading between the lines

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

  • The method assumes every client shares the same node set and adjacency; a natural extension would learn a shared spectral basis across heterogeneous graphs or align local Laplacians, which the paper leaves open.
  • The reverse-annealing schedule is linear; a history-aware aggregator (for instance, an exponential moving average of validation losses) would likely be more robust to noisy loss signals, a sensitivity the conclusion itself flags.
  • On KnowAir, where the paper reports only marginal federated gains, the correlation between validation loss and client difficulty is probably weaker; FFA's advantage would be expected to grow with the reliability of the loss signal.
  • A direct test of FFA's mechanism: perturb the validation losses with random noise on a PeMS04 split; if worst-client MAE stops improving over FedAvg, the benefit comes from the loss signal rather than from the weighting formula alone.
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 / 6 minor

Summary. This paper proposes F2STNet, a federated spatiotemporal forecasting model that combines truncated graph-Fourier spectral projection, a diagonal state-space temporal encoder, graph convolution, and a fairness-aware federated aggregation scheme (FFA). FFA reweights client updates using validation losses under a linearly increasing fairness schedule. The authors report centralized experiments on PeMS04, HZMetro, and KnowAir and federated experiments on PeMS04, with additional HZMetro and KnowAir results in an appendix, claiming consistent state-of-the-art forecasting accuracy and improved client-level fairness.

Significance. If the claims hold, the paper offers a compact architecture with sub-1.2M parameters, linear temporal complexity, and a simple, testable fairness mechanism. The model equations are internally coherent, the shared-topology assumption is explicitly stated, and the ablations and sensitivity analyses are useful for understanding the design. The main value of the contribution is the combination of known components, rather than a fundamentally new modeling principle; nevertheless, the reported efficiency numbers and the explicit client-level fairness metrics are potentially useful for practitioners. The evaluation protocol, however, currently does not provide enough statistical evidence to support the headline claims of consistent outperformance and superior fairness.

major comments (3)
  1. [§4.2 / Table 2] Section 4.4 states that the same seed list is used for all compared methods so that differences can be analyzed with paired runs, and Eq. (17) defines paired-seed means and standard deviations, but Tables 1 and 2 report standard deviations only for F2STNet, and no paired differences, confidence intervals, or tests are presented. This matters directly for the headline claim: on HZMetro in Table 1, F2STNet's RMSE is 29.213±0.15 versus 29.232 for STG-Mamba, and its MAE is 18.262±0.08 versus 18.264, so the advantage is smaller than the reported run-to-run spread. In Table 2, the Std-RMSE improvement of 0.75 versus 0.81 is a single number with no seed-to-seed variability. Please report baseline standard deviations, paired per-seed differences for the key metrics, and a significance test or effect-size measure, or explicitly soften the 'consistently outperforms SOTA baselines' claim.
  2. [§3.1 / §1] The federated baselines (FedAvg, FedProx, MOON, FedOPT, FedProc, FedSage, FedProto, FGGP) are described as 'aggregation strategies,' but the backbone model carrying these strategies is never stated. If the backbone is not F2STNet's spectral+state-space network, the accuracy and fairness gaps in Table 2 measure a mixture of architecture and aggregation effects, and the superiority of FFA over the compared aggregation methods is not established. The paper should specify the common backbone and hyperparameters used for all federated baselines, or explicitly present the comparison as full-method comparisons and include a F2STNet + vanilla FedAvg control (the 'w/o Fair Aggregation' ablation is a step in this direction but needs to be placed alongside the baselines in the same table).
  3. [§4.3 / §4.7] The problem formulation assumes a single globally shared adjacency matrix A∈R^{N×N}, and the truncated graph-Fourier basis U_F is precomputed once from this shared topology. Consequently, all clients must observe the same N nodes with the same connectivity, and the federated partition is only over time samples or records. The introduction's examples ('data silos, such as distributed traffic sensors') suggest a different and more general scenario in which clients own disjoint sensors or distinct graphs; for that scenario the spectral projection and graph convolution in Eqs. (1)–(8) are undefined. Please state the shared-topology assumption in the contributions and abstract, and discuss whether or how the method extends to disjoint-client graphs. This limitation does not invalidate the PeMS04 experiments, but it materially narrows the claimed scope of 'decentralized and heterogeneous data.'
minor comments (6)
  1. [§4.3] The sentence 'the gains on HZMetro demonstrate the benefit of fairness-aware aggregation for periodic graph structures' is confusing because the HZMetro results in Table 1 are centralized; please clarify whether this refers to the federated HZMetro results in Appendix B.
  2. [§4.2 / Table 1] The baseline name alternates between 'STAFormer' and 'STAEformer'; please standardize. Reference [30] for FedOPT is cited as ICLR 2021, but the listed source is an arXiv preprint; please correct the venue.
  3. [Figure 5] The caption says the variants are compared on all three datasets, but 'w/o Fair Aggregation' is a federated-only modification; please specify the configuration used for the centralized datasets so the reader knows what 'w/o Fair' means there.
  4. [Figure 6 / §4.7] The sensitivity analysis is described as heatmaps for λ_init, λ_max, and η, but the numerical values are not reported in the text; please include a table or numeric summary so the stability claim can be verified.
  5. [§4.7] The text reports the 'best observed configuration' without stating whether the hyperparameter search was performed on a validation split; please clarify this to rule out test-set tuning of λ_init, λ_max, η, F, and K.
  6. [All] No code or data availability statement is provided; please add a reproducibility statement or specify whether code will be released.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported predictions and fairness gains are held-out test evaluations of an independently specified model.

full rationale

F2STNet's derivation chain is self-contained. The model combines a truncated graph-Fourier projection, a diagonal state-space encoder, graph convolution, and a loss-based client reweighting rule (Eqs. 12-15). None of the reported RMSE, MAE, MAPE, Max-RMSE, or Std-RMSE values are defined as functions of the fitted parameters or of the aggregation weights; they are computed on held-out test sets and compared against external baselines. The FFA aggregation uses client validation losses as a surrogate to set aggregation weights, while the fairness claim is evaluated on test-set Max-RMSE and Std-RMSE, so the target metric is not the fitted quantity. The self-citations in the introduction and related work (e.g., [19], [20], [37], [38], [39]) motivate adjacent video-language and multimodal applications and are not load-bearing for the central forecasting or fairness claims. No uniqueness theorem or ansatz is imported from the authors' prior work. The concern that hyperparameters such as F, K, and the lambda schedule may have been tuned on the test set is a reproducibility and statistical-risk issue, not a circularity, and the paper states that checkpoints are selected using validation loss. The paper's own limitation statement in Section 6 ('sensitive to noisy validation losses and outlier clients') describes robustness risk, not a circular reduction. Overall, no load-bearing step reduces by construction to its own inputs.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new theoretical entities. Its claims rest on six fitted hyperparameters (lambda_init, lambda_max, eta, F, K, rho), a shared-graph assumption that limits applicability, a standard linear algebra fact, an empirical assumption about validation losses, and an ad hoc schedule assumption. The shared-graph premise and the validation-loss reliability premise are the most load-bearing.

free parameters (6)
  • lambda_init = 0.03
    Initial fairness weighting coefficient in Eq. (16); selected empirically via sensitivity analysis on PeMS04 worst-client MAE (Section 4.7).
  • lambda_max = 0.2
    Upper bound for the fairness schedule in Eq. (16); selected empirically (Section 4.7).
  • eta = 0.005
    Schedule slope for the reverse-annealing lambda in Eq. (16); selected empirically (Section 4.7).
  • F = 16
    Number of retained graph Fourier eigenvectors; stated as selected empirically in Section 4.4.
  • K = 10
    Number of clients in the federated partition; chosen via client-count sensitivity analysis in Appendix A.
  • rho = 0.6
    Weight in the composite utility-fairness score in Eq. (20); chosen by the authors to emphasize utility over fairness.
assumptions (4)
  • domain assumption The graph adjacency matrix A is shared and fixed across all clients, and its Laplacian eigendecomposition yields a common spectral basis U_F.
    Invoked in Section 3.1, where A is defined as shared globally and U_F is precomputed once from this topology. The spectral projection in Eq. (1) depends on this shared basis.
  • standard math The normalized graph Laplacian L has a real eigendecomposition L = U Lambda U^T with orthonormal eigenvectors.
    Standard linear algebra for symmetric matrices; used implicitly in Section 3.1 to justify the graph Fourier basis U_F.
  • domain assumption Client validation losses are a reliable signal for reweighting, so that reweighting toward higher-loss clients improves worst-client and dispersion metrics without harming global accuracy.
    Central to FFA in Section 3.3. The paper itself notes in the conclusion that the method 'remains sensitive to noisy validation losses and outlier clients'.
  • ad hoc to paper The reverse-annealing schedule lambda_t = min(lambda_max, lambda_init + eta * t) improves the utility-fairness trade-off compared to fixed or naive schedules.
    Proposed in Section 3.4 and validated only empirically on the selected datasets in Figure 7(d); no theoretical justification is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of F2STNet: Fair and Federated Spectral-Temporal Modeling for Graph Forecasting." pith.science (2026). https://pith.science/paper/5DZVS7KH

@misc{pith2026260809082,
  author       = {Pith},
  title        = {Pith review of: F2STNet: Fair and Federated Spectral-Temporal Modeling for Graph Forecasting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5DZVS7KH}},
  note         = {Machine review of arXiv:2608.09082}
}
abstract

Spatiotemporal prediction on graph-structured data is central to traffic forecasting and environmental monitoring, yet decentralized and heterogeneous data complicate both sequence modeling and collaborative training. We propose F$^2$STNet, a federated forecasting framework that combines truncated graph-Fourier features, a lightweight diagonal state-space temporal encoder, graph convolution, and Fairness-aware Federated Aggregation (FFA). The spectral branch exposes graph-frequency structure, while the state-space layer models long temporal dependencies with linear complexity in the sequence length. FFA adjusts the FedAvg prior using client validation losses and an increasing fairness schedule. Experiments on PeMS04, HZMetro, and KnowAir show favorable forecasting accuracy relative to the evaluated baselines; federated experiments on PeMS04 additionally improve worst-client and client-dispersion metrics.

Figures

Figures reproduced from arXiv: 2608.09082 by the authors.

Figure 1
Figure 1. Performance–efficiency trade-off of spatiotemporal [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. F 2 STNet Framework Overview. Our model comprises three core components: (a) a spectral projection module that decomposes each graph signal in a truncated graph Fourier basis and embeds node-level frequency contributions with a shared MLP, (b) a lightweight diagonal state-space temporal encoder for efficient sequence modeling, and (c) an adaptive fairness-aware aggregation mechanism in federated training, which dyna… view at source ↗
Figure 3
Figure 3. Graph-spectral feature extraction using the normalized [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Illustration of our fairness-aware aggregation strategy. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Performance comparison of F2 STNet variants on three datasets, showing the impact of removing spectral, temporal, and fairness-aware components. Full F2 STNet performs best across all metrics. 4.6. Ablation Study We conduct ablation experiments to assess the contributi…
Figure 7
Figure 7. Figure 7: Convergence and fairness analysis under high hetero [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Client-wise RMSE under moderate and high heterogene [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Comparison of training behavior under varying client counts ( [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Client-level training behavior comparison of different methods on HZMetro and KnowAir datasets. Each column corresponds [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 39 canonical work pages

  1. [1]

    Nested spatio-temporal time series fore- casting

    Yinghao Ai, Yukai Zhou, Ruoxi Jiang, Junyi An, Chao Qu, Zhijian Zhou, Shiyu Wang, Fenglei Cao, Zenglin Xu, Furao Shen, and Yuan Qi. Nested spatio-temporal time series fore- casting. InProceedings of the 43rd International Conference on Machine Learning, 2026. 2

    work page 2026

  2. [2]

    Spectral temporal graph neural network for multivariate time-series forecasting

    Defu Cao, Yujing Wang, Juanyong Duan, Ce Zhang, Xia Zhu, Congrui Huang, Yunhai Tong, Bixiong Xu, Jing Bai, Jie Tong, et al. Spectral temporal graph neural network for multivariate time-series forecasting. InAdvances in Neural Information Processing Systems (NeurIPS), pages 17766– 17778, 2020. 1, 2

    work page 2020

  3. [3]

    Impact of noisy supervision in foundation model learning.IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(7):5690–5707, 2025

    Hao Chen, Zihan Wang, Ran Tao, Hongxin Wei, Xing Xie, Masashi Sugiyama, Bhiksha Raj, and Jindong Wang. Impact of noisy supervision in foundation model learning.IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(7):5690–5707, 2025. 1

    work page 2025

  4. [4]

    Prompt federated learning for weather forecasting: Toward foundation models on meteorological data

    Shengchao Chen, Guodong Long, Tao Shen, and Jing Jiang. Prompt federated learning for weather forecasting: Toward foundation models on meteorological data. InProceedings of the 32nd International Joint Conference on Artificial In- telligence (IJCAI), pages 3532–3540, 2023. 2

    work page 2023

  5. [5]

    Hypercomplex prompt-aware multi- modal recommendation

    Zheyu Chen, Jinfeng Xu, Hewei Wang, Shuo Yang, Zitong Wan, and Haibo Hu. Hypercomplex prompt-aware multi- modal recommendation. InProceedings of the 34th ACM In- ternational Conference on Information and Knowledge Man- agement, pages 403–414. ACM, 2025. 2

    work page 2025

  6. [6]

    FedGCR: Achieving perfor- mance and fairness for federated learning with distinct client types via group customization and reweighting

    Shu-Ling Cheng, Chin-Yuan Yeh, Ting-An Chen, Eliana Pastor, and Min-Sheng Chen. FedGCR: Achieving perfor- mance and fairness for federated learning with distinct client types via group customization and reweighting. InProceed- ings of the AAAI Conference on Artificial Intelligence, pages 11498–11506, 2024. 2

    work page 2024

  7. [7]

    Graph neural controlled differential equa- tions for traffic forecasting

    Jeongwhan Choi, Hwangyong Choi, Jeehyun Hwang, and Noseong Park. Graph neural controlled differential equa- tions for traffic forecasting. InAAAI, 2022. 5

    work page 2022

  8. [8]

    SpoT-Mamba: Learning long-range dependency on spatio-temporal graphs with selective state spaces.arXiv preprint arXiv.2406.11244, 2024

    Jinhyeok Choi, Heehyeon Kim, Minhyeong An, and Joyce Jiyoung Whang. SpoT-Mamba: Learning long-range dependency on spatio-temporal graphs with selective state spaces.arXiv preprint arXiv.2406.11244, 2024. 5

    arXiv 2024

Show all 43 references
  1. [9]

    Ezzeldin, Shen Yan, Chaoyang He, Emilio Ferrara, and Salman A

    Yahya H. Ezzeldin, Shen Yan, Chaoyang He, Emilio Ferrara, and Salman A. Avestimehr. Fairfed: Enabling group fairness in federated learning. InProceedings of the AAAI Conference on Artificial Intelligence, pages 7494–7502, 2023. 2

    2023

  2. [10]

    Pdg2seq: Periodic dynamic graph to sequence model for traffic flow prediction.Neural Netw., 183(C), 2025

    Jin Fan, Wenchao Weng, Qikai Chen, Huifeng Wu, and Jia Wu. Pdg2seq: Periodic dynamic graph to sequence model for traffic flow prediction.Neural Netw., 183(C), 2025. 5

    2025

  3. [11]

    Mozhgan Rahmatinia, and Seyed- Amin Hosseini-Seno

    Seyed-Majid Hosseini, S. Mozhgan Rahmatinia, and Seyed- Amin Hosseini-Seno. Integrated spatio-temporal modeling with hybrid graph convolutions and the graph fourier neu- ral operator for traffic prediction.Scientific Reports, 16(1): 12945, 2026. 2

    2026

  4. [12]

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

    Jiawei Jiang, Chengkai Han, Wayne Xin Zhao, and Jingyuan Wang. Pdformer: Propagation delay-aware dynamic long- range transformer for traffic flow prediction. InAAAI. AAAI Press, 2023. 5

    2023

  5. [13]

    Graph neural network for traffic forecasting: The research progress

    Weiwei Jiang, Jiayun Luo, Miao He, and Weixi Gu. Graph neural network for traffic forecasting: The research progress. ISPRS International Journal of Geo-Information, 12(3),

  6. [14]

    Fedgraph-fair: Federated learning with per- sonalization and fairness via dynamic graphs and distribu- tionally robust optimization.Information Sciences, 728: 122710, 2026

    Koffka Khan. Fedgraph-fair: Federated learning with per- sonalization and fairness via dynamic graphs and distribu- tionally robust optimization.Information Sciences, 728: 122710, 2026. 2

    2026

  7. [15]

    Lightcts: A lightweight framework for correlated time series forecasting.Proceedings of the ACM on Management of Data, 1(2):1–26, 2023

    Zhichen Lai, Dalin Zhang, Huan Li, Christian S Jensen, Hua Lu, and Yan Zhao. Lightcts: A lightweight framework for correlated time series forecasting.Proceedings of the ACM on Management of Data, 1(2):1–26, 2023. 5

    2023

  8. [16]

    Enhancing topolog- ical dependencies in spatio-temporal graphs with cycle mes- sage passing blocks

    Minho Lee, Yun Young Choi, Sun Woo Park, Seunghwan Lee, Joohwan Ko, and Jaeyoung Hong. Enhancing topolog- ical dependencies in spatio-temporal graphs with cycle mes- sage passing blocks. InThe Third Learning on Graphs Con- ference, 2024. 2, 5

    2024

  9. [17]

    Stg-mamba: Spatial-temporal graph learning via selective state space model.arXiv preprint arXiv:2403.12418, 2024

    Lincan Li, Hanchen Wang, Wenjie Zhang, and Adelle Coster. Stg-mamba: Spatial-temporal graph learning via selective state space model.arXiv preprint arXiv:2403.12418, 2024. 2, 5

    2024 arXiv

  10. [18]

    Model- contrastive federated learning

    Qinbin Li, Bingsheng He, and Dawn Song. Model- contrastive federated learning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10713–10722, 2021. 5

    2021

  11. [19]

    Towards understanding camera motions in any video

    Zhiqiu Lin, Siyuan Cen, Daniel Jiang, Jay Karhade, Hewei Wang, Chancharik Mitra, Yu Tong Tiffany Ling, Yuhan Huang, Rushikesh Zawar, Xue Bai, Yilun Du, Chuang Gan, and Deva Ramanan. Towards understanding camera motions in any video. InAdvances in Neural Information Processing ...

    2025

  12. [20]

    Building a precise video language with human-AI oversight

    Zhiqiu Lin, Siyuan Cen, Chancharik Mitra, Isaac Li, Yuhan Huang, Yu Tong Tiffany Ling, Hewei Wang, Irene Pi, Shi- hang Zhu, Yili Han, Yilun Du, and Deva Ramanan. Building a precise video language with human-AI oversight. InPro- ceedings of the IEEE/CVF Conference on Computer V...

    2026

  13. [21]

    A general spatio-temporal backbone with scalable contextual pattern bank for urban continual forecasting

    Aoyu Liu and Yaying Zhang. A general spatio-temporal backbone with scalable contextual pattern bank for urban continual forecasting. InThe Fourteenth International Con- ference on Learning Representations, 2026. 2

    2026

  14. [22]

    Spatio-temporal adaptive embedding makes vanilla transformer sota for traf- fic forecasting

    Hangchen Liu, Zheng Dong, Renhe Jiang, Jiewen Deng, Jin- liang Deng, Quanjun Chen, and Xuan Song. Spatio-temporal adaptive embedding makes vanilla transformer sota for traf- fic forecasting. InProceedings of the 32nd ACM Interna- tional Conference on Information and Knowledge ...

    2023

  15. [23]

    Personalized fed- erated learning for spatio-temporal forecasting: A dual se- mantic alignment-based contrastive approach.arXiv preprint arXiv:2404.03702, 2024

    Qian Liu, Sheng Sun, Yi Liang, et al. Personalized fed- erated learning for spatio-temporal forecasting: A dual se- mantic alignment-based contrastive approach.arXiv preprint arXiv:2404.03702, 2024. 1

    2024 arXiv

  16. [24]

    Personalized federated learning for spatio-temporal forecasting: A dual semantic alignment-based contrastive ap- proach

    Qingxiang Liu, Sheng Sun, Yuxuan Liang, Jingjing Xue, and Min Liu. Personalized federated learning for spatio-temporal forecasting: A dual semantic alignment-based contrastive ap- proach. InProceedings of the AAAI Conference on Artificial Intelligence, pages 12192–12200, 2025. 2 9

    2025

  17. [25]

    Communication- Efficient Learning of Deep Networks from Decentralized Data

    Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, and Blaise Aguera y Arcas. Communication- Efficient Learning of Deep Networks from Decentralized Data. InProceedings of the 20th International Conference on Artificial Intelligence and Statistics, pages 1273–1282. PMLR...

    2017

  18. [26]

    Fedproc: Prototypical contrastive federated learning on non-iid data.arXiv preprint arXiv:2109.12273, 2021

    Xutong Mu, Yulong Shen, Ke Cheng, Xueli Geng, Jiaxuan Fu, Tao Zhang, and Zhiwei Zhang. Fedproc: Prototypical contrastive federated learning on non-iid data.arXiv preprint arXiv:2109.12273, 2021. 5

    2021 arXiv

  19. [27]

    Fair- ness in federated learning: Trends, challenges, and opportu- nities.Advanced Intelligent Systems, 2025

    Noorain Mukhtiar, Adnan Mahmood, and Quan Sheng. Fair- ness in federated learning: Trends, challenges, and opportu- nities.Advanced Intelligent Systems, 2025. 1

    2025

  20. [28]

    Federated spatial-temporal traffic forecasting with vmd-enhanced graph attention and lstm.Scientific Re- ports, 16(1):8852, 2026

    Tarun Mundada, Samruddhi Ramdhave, Sanyam Jain, and Saurav Gupta. Federated spatial-temporal traffic forecasting with vmd-enhanced graph attention and lstm.Scientific Re- ports, 16(1):8852, 2026. 2

    2026

  21. [29]

    Efraimidis

    Vasileios Perifanis, Nikolaos Pavlidis, Rigas-Alexandros Koutsiamanis, and Pavlos S. Efraimidis. Federated learning for 5g base station traffic forecasting.Computer Networks, 235:109950, 2023. 2

    2023

  22. [30]

    Adaptive federated optimization

    Sashank Reddi, Zachary Charles, Manzil Zaheer, Zachary Garrett, Keith Rush, Jakub Kone ˇcn`y, Sanjiv Kumar, and H Brendan McMahan. Adaptive federated optimization. arXiv e-prints, pages arXiv–2003, 2020. 5

    2003

  23. [31]

    On the convergence of federated optimization in heterogeneous networks.arXiv preprint arXiv:1812.06127, 2018

    Anit Kumar Sahu, Tian Li, Maziar Sanjabi, Manzil Zaheer, Ameet Talwalkar, and Virginia Smith. On the convergence of federated optimization in heterogeneous networks.arXiv preprint arXiv:1812.06127, 2018. 5

    2018 arXiv

  24. [32]

    Modeling multivariate biosignals with graph neural networks and structured state space

    Siyi Tang, Jared Dunnmon, Liangqiong Qu, Khaled Kamal Saab, Tina Baykaner, Christopher Lee-Messer, and Daniel Rubin. Modeling multivariate biosignals with graph neural networks and structured state space. InICLR 2023 Workshop on Time Series Representation Learning for Health, 2023. 1

    2023

  25. [33]

    Federated graph learning under domain shift with generalizable proto- types

    Guancheng Wan, Wenke Huang, and Mang Ye. Federated graph learning under domain shift with generalizable proto- types. InProceedings of the AAAI Conference on Artificial Intelligence, pages 15429–15437, 2024. 5

    2024

  26. [34]

    Unlocking dy- namic inter-client spatial dependencies: A federated spatio- temporal graph learning method for traffic flow forecasting

    Feng Wang, Tianxiang Chen, Shuyue Wei, Qian Chu, Yi Zhang, Yifan Sun, and Zhiming Zheng. Unlocking dy- namic inter-client spatial dependencies: A federated spatio- temporal graph learning method for traffic flow forecasting. InProceedings of the AAAI Conference on Artificial I...

    2026

  27. [35]

    AirShot: Efficient few-shot detection for autonomous explo- ration

    Zihan Wang, Bowen Li, Chen Wang, and Sebastian Scherer. AirShot: Efficient few-shot detection for autonomous explo- ration. In2024 IEEE/RSJ International Conference on In- telligent Robots and Systems (IROS), pages 11654–11661,

  28. [36]

    A decomposition dynamic graph con- volutional recurrent network for traffic forecasting.Pattern Recognition, page 109670, 2023

    Wenchao Weng, Jin Fan, Huifeng Wu, Yujie Hu, Hao Tian, Fu Zhu, and Jia Wu. A decomposition dynamic graph con- volutional recurrent network for traffic forecasting.Pattern Recognition, page 109670, 2023. 5

    2023

  29. [37]

    Jinfeng Xu, Zheyu Chen, Jinze Li, Shuo Yang, Hewei Wang, Yijie Li, Mengran Li, Puzhen Wu, and Edith C. H. Ngai. MDVT: Enhancing multimodal recommendation with model-agnostic multimodal-driven virtual triplets. InPro- ceedings of the 31st ACM SIGKDD Conference on Knowl- edge Di...

    2025

  30. [38]

    Jinfeng Xu, Zheyu Chen, Shuo Yang, Jinze Li, Hewei Wang, and Edith C. H. Ngai. MENTOR: Multi-level self-supervised learning for multimodal recommendation. InProceedings of the AAAI Conference on Artificial Intelligence, pages 12908– 12917, 2025. 2

    2025

  31. [39]

    Multi-cali any- thing: Dense feature multi-frame structure-from-motion for large-scale camera array calibration

    Jinjiang You, Hewei Wang, Yijie Li, Mingxiao Huo, Long Van Tran Ha, Mingyuan Ma, Jinfeng Xu, Jiayi Zhang, Puzhen Wu, Shubham Garg, and Wei Pu. Multi-cali any- thing: Dense feature multi-frame structure-from-motion for large-scale camera array calibration. In2025 IEEE/RSJ In- t...

    2025

  32. [40]

    Dual attention-based federated learning for wireless traffic prediction

    Chuanting Zhang, Shuping Dang, Basem Shihada, and Mohamed-Slim Alouini. Dual attention-based federated learning for wireless traffic prediction. InProceedings of IEEE INFOCOM, pages 1–10, 2021. 2

    2021

  33. [41]

    Subgraph federated learning with missing neighbor generation

    Ke Zhang, Carl Yang, Xiaoxiao Li, Lichao Sun, and Siu Ming Yiu. Subgraph federated learning with missing neighbor generation. InThirty-Fifth Conference on Neural Information Processing Systems, 2021. 5

    2021

  34. [42]

    Graph neural networks: A review of methods and applications.arXiv preprint arXiv:1812.08434,

    Jie Zhou, Ganqu Cui, Shengding Hu, Zhengyan Zhang, Cheng Yang, Zhiyuan Liu, Lifeng Wang, Changcheng Li, and Maosong Sun. Graph neural networks: A review of methods and applications.arXiv preprint arXiv:1812.08434,

    arXiv

  35. [43]

    Multispans: A multi-range spatial-temporal trans- former network for traffic forecast via structural entropy op- timization

    Dongcheng Zou, Senzhang Wang, Xuefeng Li, Hao Peng, Yuandong Wang, Chunyang Liu, Kehua Sheng, and Bo Zhang. Multispans: A multi-range spatial-temporal trans- former network for traffic forecast via structural entropy op- timization. InProceedings of the 17th ACM International ...

    2024

Pith tools

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