Pith. sign in

REVIEW 4 major objections 6 minor 73 references

Cellular Traffic Prediction via Deep State Space Models with Attention Mechanism

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper claims that attention-based Kalman filter models, A-LKF and A-EKF, outperform established baselines for cellular traffic prediction, with the largest gains at the one-day horizon.

desk verdict A plausible engineering combination that needs to specify how the 1-day forecasts are actually computed before the central long-term accuracy claim can be evaluated. read the letter →

arxiv 2506.15688 v1 pith:QQJ3OR4T submitted 2025-05-26 cs.LG cs.AI

Hui Ma , Kai Yang , Man-On Pun This is my paper · ORCID
classification cs.LGcs.AI
keywords cellulartrafficpredictionKalmanfilterdeepstatespacemodelspatiotemporaldependenciesattentionmechanismauxiliaryinformationlong-termforecastingextended
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

This paper tries to establish that cellular traffic forecasting can be made more accurate by combining an attention-based convolutional encoder with a Kalman filter in a single end-to-end deep state space model, and that the resulting A-LKF and A-EKF variants beat established deep learning baselines. The authors argue that the Kalman gain acts as an adaptive weight that balances each new observation against the predicted prior state, which prevents small errors from accumulating over long horizons. That property matters because operators need reliable day-ahead predictions to allocate radio resources and avoid congestion, and most sequence models degrade as the forecast horizon grows. On three real-world datasets, the paper reports that A-LKF and A-EKF achieve the best average rank among nine methods, with the largest gains at the 24-hour horizon.

What carries the argument

The load-bearing mechanism is the Kalman gain $K$, computed from the predicted covariance, the measurement matrix, and the measurement noise, and used to update the posterior state as $\hat{x}_t = K z_t + (I - K H)\hat{x}_{t|t-1}$ in A-LKF and the corresponding extended-Kalman form in A-EKF. The gain is adaptive because the observation $z_t$ and the noise covariance $R_t$ come from learned neural-network outputs, so the model can decide at every timestep how much to trust incoming traffic observations versus its own prior. Around this recursion, an attention-based CNN encoder captures spatial dependencies among neighboring cells, an autoregressive branch tracks scale changes, and an exogenous-feature branch folds in social activity and news data; all matrices in the state-space recursion are constrained to be diagonal, which keeps the computation light but restricts the dynamics the model can represent.

What would settle it

Run A-LKF and A-EKF on every cell or a random sample of cells in the Milan and Trentino grids, computing average RMSE and MAE against ST-Tran at both 1-hour and 24-hour horizons; if the proposed models do not win on average, the central claim fails. A second check is to verify the Kalman-gain visualization on synthetic data with known observation noise, where the gain should track the optimal balance between prior and observation.

Watch

Extended reading notes

Core claim

The paper's central claim is that a deep state space model in which a Kalman filter carries the temporal recursion and an attention-augmented convolutional neural network supplies the spatial representation outperforms state-of-the-art machine learning methods for cellular traffic prediction. In the authors' telling, the decisive mechanism is the Kalman gain: at each timestep it proportionally mixes the new observation, produced by the encoder from traffic and exogenous features, with the prior state, and this adaptive mixing keeps long-horizon errors from compounding. The reported experiments on Milan, Trentino, and a private LTE dataset show A-LKF and A-EKF ranking ahead of LSTM, GRU, GCN, DenseNet, DeepAuto, STCNet, and ST-Tran overall; ST-Tran is better at one-hour-ahead prediction, while A-LKF and A-EKF win at one-day-ahead prediction, which the authors attribute to error accumulation in the transformer baseline. Between the two variants, A-EKF is slightly better because its quadratic transition and measurement functions capture nonlinear traffic dynamics.

Load-bearing premise

The comparison is built on four manually chosen center cells with no stated selection criterion, so if those cells were selected because the models performed well there, the claimed advantage may not generalize to other locations.

Editorial extensions

If this is right

  • With the claimed accuracy gains at the 24-hour horizon, mobile operators could make day-ahead resource allocation and congestion management decisions from traffic forecasts rather than from short-horizon extrapolations.
  • Because the Kalman gain trajectory is visualized over a 24-hour window, the model offers a partial explanation of when the system trusts fresh observations versus its internal state, which is useful for debugging and building operator trust.
  • The paper's ablation results indicate that the autoregressive branch contributes the most to accuracy, so any successful variant of this framework should keep a linear scale-fitting component alongside the learned deep state recursion.
  • On the paper's evidence, transformer-based forecasting is strong at short horizons but loses to the Kalman-based models at one-day-ahead, suggesting the adaptive state correction is a viable remedy for long-horizon error accumulation.
  • The two variants' performance difference supports the claim that nonlinear traffic dynamics are better handled by the extended Kalman filter, pointing to nonlinear state-space models as the more promising direction for cellular traffic.

Reading between the lines

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

  • Editorial extension: the paper evaluates only four center cells chosen without stated criteria; a straightforward generalization is to apply A-LKF and A-EKF to every grid cell or a random sample, which would show whether the reported advantage is uniform or concentrated in a few favorable locations.
  • Editorial extension: the adaptive Kalman gain could be used as an online regime-change detector; a drop in the gain would indicate that observations are becoming unreliable, which operators could exploit to flag anomalous traffic patterns before they distort forecasts.
  • Editorial extension: replacing the attention-based CNN with a graph neural network over the actual base-station adjacency could preserve the Kalman recursion while using true network topology, a combination the paper does not test.
  • Editorial extension: the private LTE dataset covers only about two weeks; testing on a longer, multi-season record would reveal whether the long-horizon advantage persists across weekly and seasonal drift.
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

4 major / 6 minor

Summary. The paper proposes two deep state space models, A-LKF and A-EKF, that combine an attention-based CNN encoder, exogenous feature extraction, an autoregressive component, and a Kalman filter for cellular traffic forecasting. The models are evaluated on three real-world datasets (Milan, Trentino, and a private LTE dataset) for 1-hour and 1-day horizons, with RMSE, MAE, and CORR reported against seven baselines. The authors claim that the adaptive Kalman gain reduces error accumulation and yields superior long-term prediction accuracy, and they support this with an ablation study and a visualization of the Kalman gain over time.

Significance. If the architecture and experiments were fully specified and reproducible, the idea of making the Kalman gain a learned, adaptive balancing mechanism within an end-to-end deep forecasting model is a plausible and potentially useful contribution, especially because the Kalman filter offers a degree of interpretability that pure deep models lack. The inclusion of three datasets, including a private LTE dataset, and the ablation study are strengths. However, the paper currently leaves the central multi-step inference mechanism unspecified and provides only weak empirical evidence, so the significance of the claimed improvements cannot yet be assessed reliably.

major comments (4)
  1. [III-B, Eqs. (6)-(13)] The inference procedure for the 24-step (1-day) forecasting setting is not specified. The Kalman update at time t requires an observation z_t computed by Eq. (6) from the input x_t, but for h=24 the ground-truth inputs for future steps are unavailable. The paper never states whether the model is applied recursively with its own predictions as pseudo-observations, whether the Kalman filter is run only over the observed history followed by iterating the transition F, or whether a single forward pass directly emits 24 outputs. This is load-bearing because the claimed advantage that the Kalman gain 'adaptively balances the observation and priori state' at each timestep and alleviates accumulated errors in long-term forecasting depends on how future observations are obtained. If pseudo-observations are used, the statistical interpretation of R_t and the Kalman gain changes and the error-alleviation property would need explicit justification; if no future observations are used, the Kalman gain cannot adapt at future steps. Please specify the exact multi-step inference algorithm and, if recursive, describe how z_t is generated from predicted outputs.
  2. [Table I and Section IV-D] The empirical claim that A-LKF and A-EKF outperform state-of-the-art baselines rests on a single table with no error bars, no multiple runs, and no significance tests. The four center cells (5060, 4259, 5680, 5085) are selected with no stated criteria, and if these cells were chosen after observing performance, the conclusion may not generalize to other locations. Please report mean and standard deviation over multiple random seeds, justify the cell selection a priori or evaluate over a larger set of cells, and provide a statistical comparison (e.g., paired tests) to support the claim of superiority.
  3. [Section IV-D and Section III-B] The model is under-specified to the point of not being reproducible. The paper does not state the sequence length T, the CNN architecture (number of layers, kernel sizes, channels), the number of attention heads, the hidden dimensions D_c, D_a, D_k, D_e, or how the transition parameter γ, process noise λ, and A-EKF coefficients α0-α2, β0-β2 are produced by a neural network. These details are needed to assess the fairness of the comparison with baselines and to allow others to replicate the method. Please provide complete architectural and hyperparameter specifications.
  4. [Section V-B, Figs. 2-4] The ablation study is described only qualitatively with figures; no numerical RMSE, MAE, or CORR values are reported for A-LKF/oAr, A-LKF/oExo, and A-LKF/oAtt. The claim that removing AR has the most significant impact cannot be quantitatively verified from the figures as presented. Please report the corresponding numerical results in a table or in the text.
minor comments (6)
  1. [Table I] The average-rank values are embedded in the table without explanation; for example, the numbers '6.30', '6.10', and '2.03' are not labeled as average ranks. Please clarify the table formatting and ensure the intended bold/underline highlighting is visible.
  2. [Section IV-C and Table I] The baseline name is spelled 'DeseNet' in the baselines list but 'DenseNet' in Table I; please use a consistent spelling.
  3. [Eqs. (24)-(26)] The notation uses D both for the feature dimension and for the mean-centered variable D_i,d, which is confusing. Also, the CORR formula in Eq. (26) is ambiguous: it should be written as the product of the square roots of the two sums, i.e., sqrt(sum_d D_i,d^2) * sqrt(sum_d Dhat_i,d^2), not as a single unparenthesized expression.
  4. [Section IV-D] The phrase 'We select RMSE as the loss objective function' is imprecise: RMSE is a metric, not a loss; please state the actual training loss (e.g., mean squared error).
  5. [Section V-C and Fig. 5] The description of the Kalman gain visualization does not define exactly what is averaged ('the mean of the Kalman gain matrix is calculated and averaged over all sequences'), and no error bars are provided, so it is unclear how stable the visualized pattern is.
  6. [Section V-A] The statement that A-EKF 'slightly performs better than A-LKF' is not consistently supported by Table I, where A-LKF achieves lower RMSE in several cells and horizons; please clarify the basis for this claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found: the model is trained on training data and evaluated on held-out test data, and the Kalman gain discussion is an interpretation rather than a fitted-input prediction.

full rationale

The central claim is that A-LKF and A-EKF outperform baselines on three real-world datasets, with training data preceding the last seven days used as test data. This is a standard empirical evaluation, not a derivation that reduces to its inputs. The Kalman filter update equations (Eqs. 8-13 and 14-19) are standard textbook equations cited to [56]-[60], and the observation z_t is generated by a learned encoder from input x_t; the prediction ŷ is a fusion of learned modules. Nothing in the paper defines the target metric or test output as a function of the fitted parameters in a way that would make the reported accuracy tautological. The paper's self-citations ([5], [6], [14]) are background references in the introduction and are not load-bearing for the proposed method or its evaluation. The visualization of the Kalman gain in Fig. 5 is an interpretability exercise, not a prediction derived from a fit. The underspecified inference procedure for h=24 (whether the model is applied recursively or in a single forward pass) is a reproducibility and correctness concern, but it is not circularity: even under any of the possible inference schemes, the test labels are not used during training or evaluation. The selection of four center cells without stated criteria could affect generalization but does not make the reported comparison circular. Overall, no step in the claimed derivation chain is equivalent to its own input by construction, so the circularity score is 0.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The central claim rests on a fully learned state space: transition, noise covariances, and nonlinear functions are all fitted to training data. The most consequential free parameters are the hyperparameter choices and the selection of specific cells, neither of which is reported with sufficient detail to assess robustness.

free parameters (8)
  • Transition matrix F = diag(γ) = learned from neural network (not reported)
    Eq. (8)-(9): the state transition is assumed diagonal with entries γ learned from data; controls how the prior state propagates.
  • Process noise covariance Q = diag(λ) = learned from neural network (not reported)
    Section III-B: Q is learned via a fully connected layer; assumed diagonal.
  • Measurement noise covariance R_t = diag(l_t) = learned via FC (not reported)
    Eq. (7) and text: l_t is output of a ReLU MLP; R_t is derived from it.
  • Attention weight matrices W_Q, W_K, W_V = learned (not reported)
    Eq. (4): standard self-attention parameters.
  • A-EKF coefficients α0, α1, α2, β0, β1, β2 = learned via neural network (not reported)
    Section III-B: quadratic functions used in EKF with coefficients learned.
  • Decoder/fusion weights W_k, b_k, W_1, W_2, W_3, b = learned (not reported)
    Eq. (20)-(21): maps state and features to prediction.
  • Hyperparameters (sequence length, CNN layers, attention heads, hidden sizes) = grid search, values not reported
    Section IV-D: grid search for all tunable hyperparameters; exact values omitted.
  • Center cell selection (5060, 4259, 5680, 5085) and 25 neighbors = chosen by authors, no criteria given
    Section IV-D: only four cells from 100x100 and 117x98 grids are used; selection may affect results.
assumptions (5)
  • domain assumption Process and measurement noises are zero-mean Gaussian with known covariances.
    Section III-B: Kalman filter assumes zero-mean Gaussian noise; in practice the covariances are learned, so the 'known' part is not external.
  • ad hoc to paper State transition, process noise, and measurement noise matrices are diagonal.
    The paper assumes F, Q, R_t are diagonal (Section III-B), which is a simplification not derived from data.
  • ad hoc to paper The nonlinear functions in A-EKF are quadratic.
    The paper chooses f(x) = α0 + α1 x + α2 x^2 and h(x) = β0 + β1 x + β2 x^2, with coefficients learned; no justification for quadratic form.
  • domain assumption The 25 neighboring cells around a center cell capture all relevant spatial dependencies.
    Section IV-D: the model uses a fixed 5x5 spatial window; this assumes spatial locality.
  • domain assumption Auxiliary information (tweets, news) is temporally aligned and predictive of traffic.
    Section III-A: the model relies on external data E_t, but the paper does not analyze alignment errors or missing data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Cellular Traffic Prediction via Deep State Space Models with Attention Mechanism." pith.science (2026). https://pith.science/paper/QQJ3OR4T

@misc{pith2026250615688,
  author       = {Pith},
  title        = {Pith review of: Cellular Traffic Prediction via Deep State Space Models with Attention Mechanism},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QQJ3OR4T}},
  note         = {Machine review of arXiv:2506.15688}
}
read the original abstract

Cellular traffic prediction is of great importance for operators to manage network resources and make decisions. Traffic is highly dynamic and influenced by many exogenous factors, which would lead to the degradation of traffic prediction accuracy. This paper proposes an end-to-end framework with two variants to explicitly characterize the spatiotemporal patterns of cellular traffic among neighboring cells. It uses convolutional neural networks with an attention mechanism to capture the spatial dynamics and Kalman filter for temporal modelling. Besides, we can fully exploit the auxiliary information such as social activities to improve prediction performance. We conduct extensive experiments on three real-world datasets. The results show that our proposed models outperform the state-of-the-art machine learning techniques in terms of prediction accuracy.

Figures

Figures reproduced from arXiv: 2506.15688 by the authors.

Figure 1
Figure 1. The structure of the deep state space model. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The prediction performance in terms of RMSE. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. The prediction performance in terms of MAE. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: The prediction performance in terms of CORR. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

73 extracted references · 68 canonical work pages

  1. [1]

    Cisco annual Internet report (2018–2023) white paper,

    Cisco, “Cisco annual Internet report (2018–2023) white paper,” Tech. Rep., 2020. 8

    work page 2018

  2. [2]

    Mining the situation: Spatiotemporal traffic prediction with big data,

    J. Xu, D. Deng, U. Demiryurek, C. Shahabi, and M. van der Schaar, “Mining the situation: Spatiotemporal traffic prediction with big data,” IEEE J. Sel. Top. Sign. Proces, vol. 9, no. 4, pp. 702–715, 2015

    work page 2015

  3. [3]

    D2D communications-assisted traffic offloading in integrated cellular-WiFi networks,

    B. Feng, C. Zhang, J. Liu, and Y . Fang, “D2D communications-assisted traffic offloading in integrated cellular-WiFi networks,”IEEE Internet Things J., vol. 6, no. 5, pp. 8670–8680, 2019

    work page 2019

  4. [4]

    ITU launches new Focus Group to study machine learning in 5G systems,

    ITUNews, “ITU launches new Focus Group to study machine learning in 5G systems,” https://news.itu.int/itu-launches-new-focus-group-study- machine-learning-5G-systems/, 2017

    work page 2017

  5. [5]

    Deep network analyzer (dna): A big data analytics platform for cellular networks,

    K. Yang, R. Liu, Y . Sun, J. Yang, and X. Chen, “Deep network analyzer (dna): A big data analytics platform for cellular networks,”IEEE Internet Things J., vol. 4, no. 6, pp. 2019–2027, 2017

    work page 2019

  6. [6]

    Active learning for wireless iot intrusion detection,

    K. Yang, J. Ren, Y . Zhu, and W. Zhang, “Active learning for wireless iot intrusion detection,”IEEE Wirel. Commun., vol. 25, no. 6, pp. 19–25, 2018

    work page 2018

  7. [7]

    IoT traffic management and integration in the QoS supported network,

    B. K. J. Al-Shammari, N. Al-Aboody, and H. S. Al-Raweshidy, “IoT traffic management and integration in the QoS supported network,”IEEE Internet Things J., vol. 5, no. 1, pp. 352–370, 2018

    work page 2018

  8. [8]

    Traffic flow prediction with big data: A deep learning approach,

    Y . Lv, Y . Duan, W. Kang, Z. Li, and F. Y . Wang, “Traffic flow prediction with big data: A deep learning approach,”IEEE Trans. Intell. Transp. Syst., vol. 16, no. 2, pp. 865–873, 2015

    work page 2015

Show all 73 references
  1. [9]

    Mobile big data: The fuel for data-driven wireless,

    X. Cheng, L. Fang, L. Yang, and S. Cui, “Mobile big data: The fuel for data-driven wireless,”IEEE Internet Things J., vol. 4, no. 5, pp. 1489–1516, 2017

    2017

  2. [10]

    Large-scale mobile traffic analysis: A survey,

    D. Naboulsi, M. Fiore, S. Ribot, and R. Stanica, “Large-scale mobile traffic analysis: A survey,”IEEE Commun. Surv. Tutor ., vol. 18, no. 1, pp. 124–161, 2016

    2016

  3. [11]

    A BiGRU autoencoder remain- ing useful life prediction scheme with attention mechanism and skip connection,

    Y . Duan, H. Li, M. He, and D. Zhao, “A BiGRU autoencoder remain- ing useful life prediction scheme with attention mechanism and skip connection,”IEEE Sensors J., vol. 21, no. 9, pp. 10 905–10 914, 2021

    2021

  4. [12]

    Gated recurrent unit network-based cellular traffic prediction,

    E. P. Lens Shiang, W. Chien, C. Lai, and H. Chao, “Gated recurrent unit network-based cellular traffic prediction,” inInt. Conf. Inf. Networking, 2020

    2020

  5. [13]

    Mobile traffic prediction from raw data using LSTM networks,

    H. D. Trinh, L. Giupponi, and P. Dini, “Mobile traffic prediction from raw data using LSTM networks,” inIEEE Int. Symp. Person. Indoor Mobile Radio Commun., 2018, pp. 1827–1832

    2018

  6. [14]

    Pc2a: Predicting collective contextual anomalies via lstm with deep generative model,

    S. Dou, K. Yang, and H. V . Poor, “Pc2a: Predicting collective contextual anomalies via lstm with deep generative model,”IEEE Internet Things J., vol. 6, no. 6, pp. 9645–9655, 2019

    2019

  7. [15]

    Time-wise attention aided convolutional neural network for data-driven cellular traffic prediction,

    W. Shen, H. Zhang, S. Guo, and C. Zhang, “Time-wise attention aided convolutional neural network for data-driven cellular traffic prediction,” IEEE Wirel. Commun. Lett., vol. 10, no. 8, pp. 1747–1751, 2021

    2021

  8. [16]

    Spatiotemporal feature residual propagation for action prediction,

    H. Zhao and R. Wildes, “Spatiotemporal feature residual propagation for action prediction,” inProc. IEEE Int. Conf. Comput. Vis., 2019, pp. 7002 – 7011

    2019

  9. [17]

    Wireless traffic modeling and prediction using seasonal ARIMA models,

    Y . Shu, M. Yu, O. Yang, J. Liu, and H. Feng, “Wireless traffic modeling and prediction using seasonal ARIMA models,”IEICE Trans. Commun., vol. E88-B, no. 10, pp. 3992–3999, 2005

    2005

  10. [18]

    Traffic prediction for mobile network using Holt-Winter’s exponential smoothing,

    D. Tikunov and Toshikazu Nishimura, “Traffic prediction for mobile network using Holt-Winter’s exponential smoothing,” inInt. Conf. Softw., Telecommun. Comput. Netw., 2007, pp. 1–5

    2007

  11. [19]

    DynaNet: Neural Kalman dynamical model for motion estimation and prediction,

    C. Chen, C. X. Lu, B. Wang, N. Trigoni, and A. Markham, “DynaNet: Neural Kalman dynamical model for motion estimation and prediction,” IEEE Trans. Neural Networks Learn. Sys., vol. 32, no. 12, pp. 5479– 5491, 2021

    2021

  12. [20]

    KalmanNet: Neural network aided Kalman filtering for partially known dynamics,

    G. Revach, N. Shlezinger, X. Ni, A. L. Escoriza, R. J. G. van Sloun, and Y . C. Eldar, “KalmanNet: Neural network aided Kalman filtering for partially known dynamics,”IEEE Trans. Signal Process., vol. 70, pp. 1532–1547, 2022

    2022

  13. [21]

    Deep state space models for time series forecasting,

    S. S. Rangapuram, M. Seeger, J. Gasthaus, L. Stella, Y . Wang, and T. Januschowski, “Deep state space models for time series forecasting,” inAdv. Neural Inf. Proces. Syst., 2018, pp. 7785–7794

    2018

  14. [22]

    Time series prediction using support vector machines: A survey,

    N. I. Sapankevych and R. Sankar, “Time series prediction using support vector machines: A survey,”IEEE Comput. Intell. Mag., vol. 4, no. 2, pp. 24–38, 2009

    2009

  15. [23]

    Deep learning for network traffic monitoring and analysis (NTMA): A survey,

    M. Abbasi, A. Shahraki, and A. Taherkordi, “Deep learning for network traffic monitoring and analysis (NTMA): A survey,”Comput. Commun., vol. 170, pp. 19–41, 2021

    2021

  16. [24]

    Deep learning for intelligent wireless net- works: A comprehensive survey,

    Q. Mao, F. Hu, and Q. Hao, “Deep learning for intelligent wireless net- works: A comprehensive survey,”IEEE Commun. Surv. Tutor ., vol. 20, no. 4, pp. 2595–2621, 2018

    2018

  17. [25]

    Cellular traffic prediction with machine learning: A survey,

    W. Jiang, “Cellular traffic prediction with machine learning: A survey,” Expert Sys. Appl., vol. 201, pp. 1–19, 2022

    2022

  18. [26]

    A dual-stage attention-based recurrent neural network for time series prediction,

    Y . Qin, D. Song, H. Cheng, W. Cheng, G. Jiang, and G. W. Cottrell, “A dual-stage attention-based recurrent neural network for time series prediction,” inInt. Joint Conf. Artif. Intell., 2017, pp. 2627–2633

    2017

  19. [27]

    Spatio-temporal wireless traffic prediction with recurrent neural network,

    C. Qiu, Y . Zhang, Z. Feng, P. Zhang, and S. Cui, “Spatio-temporal wireless traffic prediction with recurrent neural network,”IEEE Wirel. Commun. Lett., vol. 7, no. 4, pp. 554–557, 2018

    2018

  20. [28]

    Short-term residential load forecasting based on LSTM recurrent neural network,

    W. Kong, Z. Y . Dong, Y . Jia, D. J. Hill, Y . Xu, and Y . Zhang, “Short-term residential load forecasting based on LSTM recurrent neural network,” IEEE Trans. Smart Grid, vol. 10, no. 1, pp. 841–851, 2019

    2019

  21. [29]

    A deep learning framework with spatial-temporal attention mechanism for cellular traffic prediction,

    Y . Gao, X. Wei, L. Zhou, and H. Lv, “A deep learning framework with spatial-temporal attention mechanism for cellular traffic prediction,” in IEEE Globecom Workshops, 2019, pp. 1–6

    2019

  22. [30]

    Cellular traffic load prediction with LSTM and Gaussian process regression,

    W. Wang, C. Zhou, H. He, W. Wu, and X. S. Shen, “Cellular traffic load prediction with LSTM and Gaussian process regression,” inIEEE Int. Conf. Commun., 2020, pp. 1–6

    2020

  23. [31]

    Data-augmentation- based cellular traffic prediction in edge-computing-enabled smart city,

    Z. Wang, J. Hu, G. Min, Z. Zhao, and J. Wang, “Data-augmentation- based cellular traffic prediction in edge-computing-enabled smart city,” IEEE Trans. Ind. Inf., vol. 17, no. 6, pp. 4179–4187, 2021

    2021

  24. [32]

    Mobile demand forecasting via deep graph-sequence spatiotemporal modeling in cellular networks,

    L. Fang, X. Cheng, H. Wang, and L. Yang, “Mobile demand forecasting via deep graph-sequence spatiotemporal modeling in cellular networks,” IEEE Internet Things J., vol. 5, no. 4, pp. 3091–3101, 2018

    2018

  25. [33]

    Cellular network traffic prediction incorporating handover: A graph convolutional approach,

    S. Zhao, X. Jiang, G. Jacobson, R. Jana, W.-L. Hsu, R. Rustamov, M. Ta- lasila, S. A. Aftab, Y . Chen, and C. Borcea, “Cellular network traffic prediction incorporating handover: A graph convolutional approach,” inIEEE Commun. Soc. Conf. Sens., Mesh Ad Hoc Commun. Netw. worksh...

    2020

  26. [34]

    Graph attention spatial- temporal network with collaborative global-local learning for citywide mobile traffic prediction,

    K. He, X. Chen, Q. Wu, S. Yu, and Z. Zhou, “Graph attention spatial- temporal network with collaborative global-local learning for citywide mobile traffic prediction,”IEEE Trans. Mob. Comput., vol. 21, no. 4, pp. 1244–1256, 2022

    2022

  27. [35]

    Graph-based deep learning for communication networks: A survey,

    W. Jiang, “Graph-based deep learning for communication networks: A survey,”Comput. Commun., vol. 185, pp. 40–54, 2022

    2022

  28. [36]

    Network traffic prediction based on diffusion convo- lutional recurrent neural networks,

    D. Andreoletti, S. Troia, F. Musumeci, S. Giordano, G. Maier, and M. Tornatore, “Network traffic prediction based on diffusion convo- lutional recurrent neural networks,” inIEEE Conf. Comput. Commun. Workshops, 2019, pp. 246–251

    2019

  29. [37]

    Big data driven mobile traffic understanding and forecasting: A time series approach,

    F. Xu, Y . Lin, J. Huang, D. Wu, H. Shi, J. Song, and Y . Li, “Big data driven mobile traffic understanding and forecasting: A time series approach,”IEEE Trans. Serv. Comput., vol. 9, no. 5, pp. 796–805, 2016

    2016

  30. [38]

    Spatio-temporal analysis and prediction of cellular traffic in metropo- lis,

    X. Wang, Z. Zhou, F. Xiao, K. Xing, Z. Yang, Y . Liu, and C. Peng, “Spatio-temporal analysis and prediction of cellular traffic in metropo- lis,”IEEE Trans. Mob. Comput., vol. 18, no. 9, pp. 2190–2202, 2019

    2019

  31. [39]

    Towards accurate prediction for high-dimensional and highly-variable cloud workloads with deep learning,

    Z. Chen, J. Hu, G. Min, A. Y . Zomaya, and T. El-Ghazawi, “Towards accurate prediction for high-dimensional and highly-variable cloud workloads with deep learning,”IEEE Trans. Parallel Distrib. Syst., vol. 31, no. 4, pp. 923–934, 2020

    2020

  32. [40]

    Spatiotemporal modeling and prediction in cellular networks: A big data enabled deep learning approach,

    W. Jing, T. Jian, Z. Xu, Y . Wang, and D. Yang, “Spatiotemporal modeling and prediction in cellular networks: A big data enabled deep learning approach,” inProc. IEEE Conf. Comput. Commun., 2017, pp. 1–9

    2017

  33. [41]

    Citywide cellular traffic prediction based on densely connected convolutional neural networks,

    C. Zhang, H. Zhang, D. Yuan, and M. Zhang, “Citywide cellular traffic prediction based on densely connected convolutional neural networks,” IEEE Commun. Lett., vol. 22, no. 8, pp. 1656–1659, 2018

    2018

  34. [42]

    DeepRTP: A deep spatio-temporal residual network for regional traffic prediction,

    Z. Liu, M. Huang, Z. Ye, and K. Wu, “DeepRTP: A deep spatio-temporal residual network for regional traffic prediction,” inInt. Conf. Mob. Ad- Hoc Sens. Networks, 2019, pp. 291–296

    2019

  35. [43]

    A study of deep learning networks on mobile traffic forecasting,

    C. W. Huang, C. T. Chiang, and Q. Li, “A study of deep learning networks on mobile traffic forecasting,” inIEEE Int. Symp. Person. Indoor Mobile Radio Commun., 2017, pp. 1–6

    2017

  36. [44]

    Deep irregular convolutional residual LSTM for urban traffic passenger flows prediction,

    B. Du, H. Peng, S. Wang, M. Z. A. Bhuiyan, L. Wang, Q. Gong, L. Liu, and J. Li, “Deep irregular convolutional residual LSTM for urban traffic passenger flows prediction,”IEEE Trans. Intell. Transp. Syst., vol. 21, no. 3, pp. 972–985, 2020

    2020

  37. [45]

    Spatial-temporal attention-convolution network for citywide cellular traffic prediction,

    N. Zhao, Z. Ye, Y . Pei, Y .-C. Liang, and D. Niyato, “Spatial-temporal attention-convolution network for citywide cellular traffic prediction,” IEEE Commun. Lett., vol. 24, no. 11, pp. 2532–2536, 2020

    2020

  38. [46]

    ST-Tran: Spatial-temporal transformer for cellular traffic prediction,

    Q. Liu, J. Li, and Z. Lu, “ST-Tran: Spatial-temporal transformer for cellular traffic prediction,”IEEE Commun. Lett., vol. 25, no. 10, pp. 3325–3329, 2021

    2021

  39. [47]

    Deep learning for spatio-temporal data mining: A survey,

    S. Wang, J. Cao, and P. Yu, “Deep learning for spatio-temporal data mining: A survey,”IEEE Trans. Knowl. Data Eng., vol. 34, no. 8, pp. 3681–3700, 2020

    2020

  40. [48]

    DeepTP: An end-to- end neural network for mobile cellular traffic prediction,

    J. Feng, X. Chen, R. Gao, M. Zeng, and Y . Li, “DeepTP: An end-to- end neural network for mobile cellular traffic prediction,”IEEE Network, vol. 32, no. 6, pp. 108–115, 2018

    2018

  41. [49]

    ST-DenNetFus: A new deep learning approach for network demand prediction,

    H. Assem, B. Caglayan, T. S. Buda, and D. O’Sullivan, “ST-DenNetFus: A new deep learning approach for network demand prediction,” inJoint Eur . Conf. Mach. Learn. Knowl. Disc. Databases, vol. 11053, 2018, pp. 222–237. 9

    2018

  42. [50]

    Deep transfer learning for intelligent cellular traffic prediction based on cross-domain big data,

    C. Zhang, H. Zhang, J. Qiao, D. Yuan, and M. Zhang, “Deep transfer learning for intelligent cellular traffic prediction based on cross-domain big data,”IEEE J. Sel. Areas Commun., vol. 37, no. 6, pp. 1389–1401, 2019

    2019

  43. [51]

    DeepAuto: A hierarchical deep learning framework for real-time prediction in cellular networks

    A. Bhorkar, K. Zhang, and J. Wang, “DeepAuto: A hierarchical deep learning framework for real-time prediction in cellular networks.”arXiv preprint arXiv:2001.01553, 2019

    2001 arXiv

  44. [52]

    From Twitter to traffic predictor: Next-day morning traffic prediction using social media data,

    W. Yao and S. Qian, “From Twitter to traffic predictor: Next-day morning traffic prediction using social media data,”Transp. Res. Part C Emerg. Technol., vol. 124, pp. 1–26, 2021

    2021

  45. [53]

    Modeling long- and short- term temporal patterns with deep neural networks,

    G. Lai, W.-C. Chang, Y . Yang, and H. Liu, “Modeling long- and short- term temporal patterns with deep neural networks,” inInt. ACM SIGIR Conf. Res. Dev. Inf. Retr ., 2018, pp. 95–104

    2018

  46. [54]

    Towards better forecasting by fusing near and distant future visions,

    J. Cheng, K. Huang, and Z. Zheng, “Towards better forecasting by fusing near and distant future visions,” inAAAI Conf. Artif. Intell., 2020

    2020

  47. [55]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, u. Kaiser, and I. Polosukhin, “Attention is all you need,” inAdv. Neural Inf. Proces. Syst., 2017, pp. 6000–6010

    2017

  48. [56]

    Recurrent Kalman networks: Factorized inference in high-dimensional deep feature spaces,

    P. Becker, H. Pandya, G. Gebhardt, C. Zhao, J. Taylor, and G. Neumann, “Recurrent Kalman networks: Factorized inference in high-dimensional deep feature spaces,” inInt. Conf. Machin. Learn., 2019, pp. 544–552

    2019

  49. [57]

    Long short-term memory Kalman filters: Recurrent neural estimators for pose regularization,

    H. Coskun, F. Achilles, R. Dipietro, N. Navab, and F. Tombari, “Long short-term memory Kalman filters: Recurrent neural estimators for pose regularization,” inIEEE Int. Conf. Comput. Vis., 2017, pp. 5525–5533

    2017

  50. [58]

    Moving horizon estimation for multirate system with time-varying time-delays,

    Y . Gu, Y . Chou, J. Liu, and Y . Ji, “Moving horizon estimation for multirate system with time-varying time-delays,”J. Franklin Inst., vol. 356, no. 4, pp. 2325–2345, 2019

    2019

  51. [59]

    Kalman filtering in R,

    T. Fernando, “Kalman filtering in R,”J. Stat. Softw., vol. 39, no. 2, pp. 1–27, 2011

    2011

  52. [60]

    Stochastic stability of the discrete-time extended Kalman filter,

    K. Reif, S. Gunther, E. Yaz, and R. Unbehauen, “Stochastic stability of the discrete-time extended Kalman filter,”IEEE Trans. Autom. Control, vol. 44, no. 4, pp. 714–728, 1999

    1999

  53. [61]

    Kalman filters for nonlinear systems and heavy-tailed noise,

    M. Roth, “Kalman filters for nonlinear systems and heavy-tailed noise,” Ph.D. dissertation, Link ¨oping University Electronic Press, 2013

    2013

  54. [62]

    On iterative unscented Kalman filter using optimization,

    M. A. Skoglund, F. Gustafsson, and G. Hendeby, “On iterative unscented Kalman filter using optimization,” inInt. Conf. Inf. Fusion, 2019, pp. 1–8

    2019

  55. [63]

    A multi-source dataset of urban life in the city of Milan and the province of Trentino,

    G. Barlacchi, M. De Nadai, R. Larcher, A. Casella, C. Chitic, G. Torrisi, F. Antonelli, A. Vespignani, A. Pentland, and B. Lepri, “A multi-source dataset of urban life in the city of Milan and the province of Trentino,” Sci. Data, vol. 2, no. 1, pp. 1–15, 2015

    2015

  56. [64]

    Telecommunications - SMS, Call, Internet - MI,

    T. Italia, “Telecommunications - SMS, Call, Internet - MI,” 2015. [Online]. Available: https://doi.org/10.7910/DVN/EGZHFV

    2015 doi

  57. [65]

    Telecommunications - SMS, Call, Internet - TN,

    T. Italia, “Telecommunications - SMS, Call, Internet - TN,” 2015. [Online]. Available: https://doi.org/10.7910/DVN/QLCABU

    2015 doi

  58. [66]

    Social Pulse - Milano,

    SpazioDati and D. P. di Milano, “Social Pulse - Milano,” 2015. [Online]. Available: https://doi.org/10.7910/DVN/9IZALB

    2015 doi

  59. [67]

    MilanoToday,

    Citynews, “MilanoToday,” 2015. [Online]. Available: https://doi.org/10. 7910/DVN/QWOE1R

    2015

  60. [68]

    Social Pulse - Trentino,

    SpazioDati, “Social Pulse - Trentino,” 2015. [Online]. Available: https://doi.org/10.7910/DVN/5H0NUI

    2015 doi

  61. [69]

    TrentoToday,

    Citynews, “TrentoToday,” 2015. [Online]. Available: https://doi.org/10. 7910/DVN/NYQ23N

    2015

  62. [70]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” inInt. Conf. Learn. Represent., 2015

    2015

  63. [71]

    Entry-flipped transformer for inference and prediction of participant behavior,

    H. Bo and C. Tat-Jen, “Entry-flipped transformer for inference and prediction of participant behavior,” inEur . Conf. Comput. Vis., 2022

    2022

  64. [72]

    Interpretable deep feature propagation for early action recognition,

    H. Zhao and R. P. Wildes, “Interpretable deep feature propagation for early action recognition,”arXiv preprint arXiv: 2107.05122, 2021. Hui Mareceived her B.Eng. degree in 2014 and M.S. degree in 2017 from Jiangnan University, Wuxi, China. She is a Ph.D. candidate in the De- p...

    2021 arXiv

  65. [2008]

    Prior to joining CUHKSZ in 2015, he held research positions at Huawei (USA), Mitsubishi Electric Research Labs (MERL), Boston and Sony, Tokyo, Japan

    He is currently an Associate Professor with the School of Science and Engineering, The Chinese University of Hong Kong at Shenzhen (CUHKSZ), Shenzhen. Prior to joining CUHKSZ in 2015, he held research positions at Huawei (USA), Mitsubishi Electric Research Labs (MERL), Boston ...

    2015

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.