Pith. sign in

REVIEW 3 major objections 5 minor 60 references

GKNet: Graph Kalman Filtering and Model Inference via Model-based Deep Learning

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read This paper proposes a graph state-space model in which a latent signal diffuses over the graph with edge-level noise and is observed through a graph filter, and it argues that both the dynamics and the observation model can be learned…

desk verdict A useful architecture and honest experiments, but the central discretization is wrong and contradicts the paper's own Remark 1; send to referees with a demand to fix and rerun. read the letter →

arxiv 2506.22004 v1 pith:Q5XN2QBT submitted 2025-06-27 cs.LG

classification cs.LG
keywords graphsignalprocessingstatespacemodelsKalmanfiltermodel-baseddeeplearningtimeseriesforecastingneuralnetworksdataefficiencyuncertaingraphs
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper is trying to establish that a graph time series can be modeled as a latent diffusion process with edge-level noise, observed through a graph filter, and that both the dynamics and the observation model can be learned from partial measurements. The authors first solve this with an expectation-maximization algorithm wrapped around a graph Kalman smoother, then unroll the same recursions into a deep architecture, GKNet, that replaces the analytic Kalman gain with learned graph filters driven by a recurrent inference module. The payoff would be a forecasting and imputation method that stays accurate when training data are scarce and when the graph is only approximately known, as the stormwater case study argues. If right, this gives a principled way to inject graph structure into deep time-series models without requiring large training sets.

What carries the argument

The load-bearing object is the graph state-space model, whose state transition is the graph Laplacian $L$ acting as a diffusion operator, whose noise is dispersed along edges through the node-to-edge incidence matrix $B$ with learnable per-edge strengths $\alpha$, giving covariance $Q = B\operatorname{diag}(\alpha^2)B^\top$, and whose observation operator is a masked graph filter $H = M\tilde{H}(L)$ of order $K$. In GKNet, the same structure is unrolled: the prediction step uses $-cL$ and $Q$, the correction step uses a graph filter $\hat{K}_t(L)$ in place of the Kalman gain, and a recurrent inference module outputs the filter coefficients, diffusivity $c$, and time-varying edge uncertainties from encoder statistics.

What would settle it

Simulate a known diffusion $dx_t = -cLx_t dt + B\operatorname{diag}(\alpha)d\beta_t$ on a small graph, fit the proposed EM and GKNet to one-step transitions, and compare the estimated transition matrix against both $I - cL$ and $-cL$; the correct discrete model is the one that matches the simulated conditional mean $E[x_{t+1}|x_t]$, and a mismatch would falsify the state equation used in Equations (7), (17), and (35).

Watch

Extended reading notes

Core claim

The central claim is that the pair consisting of a graph-diffusion state equation and a graph-filtered observation equation is identifiable and learnable for graph time series. The state evolves as a stochastic heat diffusion $dx_t = -cLx_t dt + B\operatorname{diag}(\alpha)d\beta_t$, with per-edge uncertainty $\alpha$; observations are $y_t = M\tilde{H}(L)x_t + v_t$, a sampled graph-filtered version of the state. The paper argues that this parameterization has few parameters, that the maximum-likelihood and EM route is the principled baseline, and that unrolling the graph Kalman recursions in GKNet, replacing the Kalman gain by a learned graph filter and the model parameters by an RNN hidden state, yields a scalable architecture that learns state and observation parameters jointly and end to end. In their experiments, the claim is that this design keeps forecasting and interpolation errors low when training data are scarce and the graph is noisy.

Load-bearing premise

Everything rests on the discrete transition $x_{t+1} = -cLx_t + B\operatorname{diag}(\alpha)w_t$ being a faithful Euler discretization of the graph diffusion; if the step should read $x_{t+1} = (I - cL)x_t + \cdots$, then the Kalman prediction, likelihood, and loss all change.

Editorial extensions

If this is right

  • If the state-space model is correctly learned, missing node values can be imputed from partial observations because the diffusion and the graph filter constrain what an unobserved node should do.
  • When training data are limited, the model's inductive bias, a graph diffusion plus edge-local noise, lets it forecast and interpolate in regimes where fully data-driven neural networks diverge.
  • The learned edge uncertainties $\alpha$ provide a per-edge measure of where the assumed graph differs from the true process, so the same framework can flag unreliable connections.
  • Because GKNet's inference module has size governed by the filter order rather than the node count, the Kalman-gain and uncertainty filters scale to larger graphs than the $O(TN^3)$ EM route.

Reading between the lines

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

  • The per-edge uncertainties suggest a direct diagnostic use: train the model on a graph whose edges have been artificially perturbed and check whether the learned $\alpha$ values concentrate on those edges, which would validate the uncertainty interpretation.
  • Because the Kalman-gain filter order is independent of graph size, the same recursions could be applied to a differently sized graph after retraining the encoder and decoder; the water-network transfer experiment is a partial demonstration.
  • The runoff-driven extension indicates the framework can handle exogenous graph inputs, so a natural next step is multi-step ahead forecasting by feeding predicted states back through the learned input filter.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes a graph state-space model for time-varying graph signals, with a latent process driven by a graph SPDE with edge-localized noise and observations given by a masked graph filter. The authors first learn parameters and latent states via an EM algorithm based on Kalman smoothing, then extend this into GKNet, a model-based deep network that replaces the Kalman gain with a graph filter and estimates the model parameters with a recurrent inference module. The method is evaluated on traffic forecasting, weather interpolation, synthetic tracking, and a stormwater-network case study, with emphasis on low-data regimes.

Significance. If the theoretical derivation were correct, the paper would offer a useful model-based deep learning approach combining interpretability with scalability, and the extensive experiments, especially the water-network case study, would be valuable. The attempt to connect an SPDE-based graph kernel to recursive Kalman inference is a worthwhile direction, and the low-data empirical results are potentially interesting. However, the load-bearing derivations contain fundamental errors, so the significance is not realized in the current form: the empirical results cannot be attributed to the claimed SPDE state-space model without correcting and re-running the core derivation.

major comments (3)
  1. [II-A, Eq. (7)] Equation (7) asserts that x_{t+1} = -cL x_t + Bdiag(alpha)w_t is the first-order Euler discretization of d x_t = -cL x_t dt + Bdiag(alpha)d beta_t. Explicit Euler with unit step gives x_{t+1} = x_t - cL x_t + ... = (I - cL)x_t + ..., so the identity term is missing. This is not a typo: the same recurrence appears in the state-space model (12), the negative log-likelihood (16), the Kalman prediction (17), the smoother gains (19)-(22), the GKNet loss (35), and the water-network state equation (39). It also contradicts Remark 1, where the continuous mean is exp(-cL t)mu0; the discrete model has mean (-cL)^t mu0, which neither approximates the heat semigroup nor is stable for general c. Since the paper's central claim is that GKNet is a principled unrolling of a graph Kalman filter derived from this SPDE, the derivation does not support the claim as written.
  2. [III-B, Eqs. (19)-(23)] The EM derivation contains additional load-bearing errors. In the RTS-type smoother, Eq. (21) should read \hat P_{t-1} = P^+_{t-1} + J_{t-1}(\hat P_t - P^-_t)J^T_{t-1}, but the manuscript uses P^+_{t-1} instead of the predicted covariance P^-_t. More importantly, Eq. (23) uses \hat P_t as if it were the second moment E[x_t x_t^T], whereas throughout the paper \hat P_t denotes the smoothed covariance Cov[x_t|y]. The quadratic terms in (16) require second moments, so the M-step objective is missing the \hat x_t \hat x_t^T and \hat x_{t-1} \hat x_{t-1}^T contributions. The resulting EM parameter updates are therefore not maximum-likelihood updates for the stated model.
  3. [IV-B, Eqs. (28)-(32)] The correction module is internally inconsistent. Equation (28) defines the update as xt = x^-_t + \hat K_t(\hat x_t - x^-_t) with \hat K_t a graph filter, and Eq. (29) sets P^+_t = (I - \hat K_t)P^-_t. But the justification preceding these equations, Eqs. (30)-(31), shows that the object being approximated is P^-_t H^T(HP^-_t H^T + sigma^2 I)^{-1} H, which is the product of the Kalman gain with H, not the Kalman gain itself, and it still contains the P^-_t factor. The covariance update in Eq. (32) is also different from Eq. (29), containing P^-_t \hat K_t rather than \hat K_t. Moreover, the claim that this can be represented as a graph filter is asserted 'up to the eigenspace mismatch between L and P^-_t' without any argument that P^-_t approximately commutes with L or is polynomial in L. Without such justification, the correction sub-module is not a principled unrolling of the Kalman correction step.
minor comments (5)
  1. [IV-C, Eq. (33)] The text states that sigma(·) is the sigmoid activation, but the update for z_t in Eq. (33) uses ReLU, and the observation-noise vector sigma_t from Eq. (25) shares the same symbol as the sigmoid function. Please clarify the intended activation and notation.
  2. [III-B, Eq. (23)] The symbol Q is reused for both the expected negative log-likelihood and the system noise covariance Bdiag(alpha^2)B^T; different letters should be used to avoid confusion.
  3. [V-A] The text says 'codes are available here' but no URL or repository is provided; a working link is needed.
  4. [VI] There are several typos, including 'for for' and 'becasue' in the water-network section and 'comlexity' in Section IV-C; these should be corrected.
  5. [V] Although the standard deviations are reported as O(10^-3), Tables I-III show no error bars or confidence intervals; given the small performance differences among deep models, this makes it difficult to judge which differences are significant.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: GKNet's parameters are learned from training data and evaluated on held-out data, and the SPDE-to-state-space derivation is self-contained in the text.

full rationale

The paper's central claim is that GKNet unrolls a graph Kalman filter derived from a graph-SPDE state model and learns state/observation parameters end-to-end. No prediction in the paper reduces by construction to a fitted value. The model parameters (c, alpha, h, sigma^2) are estimated from training data via EM or the GKNet loss and then evaluated on held-out test sets, so the empirical claims are externally grounded. The graph-SPDE model is introduced and derived in the paper itself (Eqs. 4-7), with the SPDE kernel idea attributed to the external reference [25]; the Kalman recursions and EM algorithm are standard and cited to external references [46,47,49]. The self-citations ([1], [14], [31]) appear as provenance or related-work context, not as the load-bearing justification for the derivation. The reviewer-identified issue that Eq. (7) is not the correct Euler discretization of Eq. (6) — missing the identity term — is a mathematical correctness concern, not a circularity: it does not make a predicted quantity equivalent to an input by construction. Similarly, the latent uncertainty parameter alpha is learned from data and could absorb residuals, but this is a modeling flexibility issue, not circularity. Overall, the derivation chain is self-contained and the performance claims are benchmarked against external baselines, so no significant circularity is present.

Assumptions & free parameters 7 free parameters · 6 assumptions · 2 invented entities

The model rests on a parametric graph state-space formulation: a scalar diffusivity, per-edge noise dispersions, small-order graph filters, and (in GKNet) encoder/decoder and recurrent weights. The most fragile input is the claim that Eq. (7) is an Euler discretization of the SPDE, which is false; this is treated as an ad hoc assumption in the ledger. The graph-filter approximation of the Kalman gain is also asserted without a formal guarantee.

free parameters (7)
  • Diffusivity c = not reported
    Scales the graph Laplacian in the state transition (Eqs. 7, 17, 35); learned from data in both EM and GKNet.
  • Edge uncertainty vector alpha = not reported
    Per-edge noise dispersion in Q = B diag(alpha^2) B^T; in GKNet generated by an edge filter with learned coefficients (Eq. 34).
  • Observation filter coefficients h = not reported
    Graph filter H(L) = sum h_k L^k in Eq. 3; learned by EM M-step and by the GKNet inference module.
  • Observation noise variance sigma^2 = not reported
    Isotropic covariance in Eq. 15; learned in the EM algorithm.
  • Kalman gain filter coefficients hkt = not reported
    GKNet replaces the Kalman gain with graph filter Kt(L,sigma^2) whose coefficients are outputs of the recurrent inference module (Section IV-B).
  • Encoder/decoder GCNN and RNN weights = not reported
    All trainable network parameters in Eqs. 24-26 and 33, optimized by backpropagation.
  • Regularization weight lambda = 0.05, 0.15, 0.025
    Hand-chosen per experiment via validation; balances data fidelity and the state-dynamics penalty in losses (35) and (42).
assumptions (6)
  • standard math Combinatorial graph Laplacian L = BB^T and edge Laplacian L1 = B^T B
    Background definitions for graph filters and SPDEs in Section II.
  • domain assumption The data follow the linear Gaussian state-space model (12) with zero-mean Gaussian initial state, process noise, and observation noise
    Basis for Kalman smoothing and EM in Sections III and IV.
  • ad hoc to paper The discrete transition x_{t+1} = -cLx_t is a valid Euler discretization of the graph SPDE (6)
    Eq. (7); this is false (the identity term is missing) and it is load-bearing for Eq. (17), Eq. (16), Eq. (35), and Eq. (39).
  • ad hoc to paper The Kalman gain can be approximated by a graph filter Kt(L,sigma^2) despite the eigenspace mismatch with P^-_t
    Eqs. (30)-(31) in Section IV-B; asserted without a formal error bound.
  • ad hoc to paper Edge uncertainties alpha_t are generated by an edge-convolutional filter H_e(L1) of fixed order K
    Eq. (34); restricts the uncertainty model to a fixed filter order independent of graph size.
  • domain assumption GCNN encoders/decoders can map observations into a latent space where the linear transition -cL is accurate
    Central inductive bias of GKNet; validated only empirically.
invented entities (2)
  • Edge-wise uncertainty vector alpha
    purpose: Allows per-edge deviations from graph heat diffusion in the latent state; adds degrees of freedom to Eq. (6)-(7).
    A latent modeling construct learned from data; no external falsifiable prediction is attached to alpha.
  • Latent graph state x_t
    purpose: Unobserved state driving observations through the graph filter H(L); enables Kalman filtering and GKNet.
    Standard state-space latent variable; no independent physical evidence claimed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GKNet: Graph Kalman Filtering and Model Inference via Model-based Deep Learning." pith.science (2026). https://pith.science/paper/Q5XN2QBT

@misc{pith2026250622004,
  author       = {Pith},
  title        = {Pith review of: GKNet: Graph Kalman Filtering and Model Inference via Model-based Deep Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Q5XN2QBT}},
  note         = {Machine review of arXiv:2506.22004}
}
read the original abstract

Inference tasks with time series over graphs are of importance in applications such as urban water networks, economics, and networked neuroscience. Addressing these tasks typically relies on identifying a computationally affordable model that jointly captures the graph-temporal patterns of the data. In this work, we propose a graph-aware state space model for graph time series, where both the latent state and the observation equation are parametric graph-induced models with a limited number of parameters that need to be learned. More specifically, we consider the state equation to follow a stochastic partial differential equation driven by noise over the graphs edges accounting not only for potential edge uncertainties but also for increasing the degrees of freedom in the latter in a tractable manner. The graph structure conditioning of the noise dispersion allows the state variable to deviate from the stochastic process in certain neighborhoods. The observation model is a sampled and graph-filtered version of the state capturing multi-hop neighboring influence. The goal is to learn the parameters in both state and observation models from the partially observed data for downstream tasks such as prediction and imputation. The model is inferred first through a maximum likelihood approach that provides theoretical tractability but is limited in expressivity and scalability. To improve on the latter, we use the state-space formulation to build a principled deep learning architecture that jointly learns the parameters and tracks the state in an end-to-end manner in the spirit of Kalman neural networks.

Figures

Figures reproduced from arXiv: 2506.22004 by the authors.

Figure 1
Figure 1. Recurrent Graph Kalman Network block diagram. Encoder: takes the observations as input and transfer them into the [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 3
Figure 3. The performance of data-driven models on weather [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. The performance of data-driven models on weather [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figures from the paper (3 more)
Figure 5
Figure 5. Figure 5: The performance of data-driven models on tracking [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 7
Figure 7. Figure 7: The performance of the GKNet model in predicting the [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: The performance of the GKNet model in predicting the [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

60 extracted references · 52 canonical work pages

  1. [1]

    Inferring time varying signals over uncertain graphs,

    M. Sabbaqi and E. Isufi, “Inferring time varying signals over uncertain graphs,” in ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , 2024, pp. 9876– 9880

  2. [2]

    Statistical models for network graphs,

    E. D. Kolaczyk, G. Cs ´ardi, E. D. Kolaczyk, and G. Cs ´ardi, “Statistical models for network graphs,” Statistical analysis of network data with R , pp. 85–109, 2014

  3. [3]

    Graph signal processing: Overview, challenges, and appli- cations,

    A. Ortega, P. Frossard, J. Kova ˇcevi´c, J. M. F. Moura, and P. Van- dergheynst, “Graph signal processing: Overview, challenges, and appli- cations,” Proceedings of the IEEE , vol. 106, no. 5, pp. 808–828, 2018

  4. [4]

    Learning to reconstruct missing data from spatiotemporal graphs with sparse observations,

    I. Marisca, A. Cini, and C. Alippi, “Learning to reconstruct missing data from spatiotemporal graphs with sparse observations,” in Advances in Neural Information Processing Systems , vol. 35. Curran Associates, Inc., 2022, pp. 32 069–32 082

  5. [5]

    Fouriergnn: Rethinking multivariate time series forecasting from a pure graph perspective,

    K. Yi, Q. Zhang, W. Fan, H. He, L. Hu, P. Wang, N. An, L. Cao, and Z. Niu, “Fouriergnn: Rethinking multivariate time series forecasting from a pure graph perspective,” in Advances in Neural Information Processing Systems, vol. 36. Curran Associates, Inc., 2023, pp. 69 638– 69 660

  6. [6]

    Deep learning for anomaly detection in multivariate time series: Approaches, applications, and challenges,

    G. Li and J. J. Jung, “Deep learning for anomaly detection in multivariate time series: Approaches, applications, and challenges,” Information Fusion, vol. 91, pp. 93–102, 2023

  7. [7]

    Topology identi- fication and learning over graphs: Accounting for nonlinearities and dynamics,

    G. B. Giannakis, Y . Shen, and G. V . Karanikolas, “Topology identi- fication and learning over graphs: Accounting for nonlinearities and dynamics,” Proceedings of the IEEE, vol. 106, no. 5, pp. 787–807, 2018

  8. [8]

    Learning graphs from data: A signal representation perspective,

    X. Dong, D. Thanou, M. Rabbat, and P. Frossard, “Learning graphs from data: A signal representation perspective,” IEEE Signal Processing Magazine, vol. 36, no. 3, pp. 44–63, 2019

Show all 60 references
  1. [9]

    Connecting the dots: Identifying network structure via graph signal processing,

    G. Mateos, S. Segarra, A. G. Marques, and A. Ribeiro, “Connecting the dots: Identifying network structure via graph signal processing,” IEEE Signal Processing Magazine , vol. 36, no. 3, pp. 16–43, 2019

  2. [10]

    Learning graphs from smooth and graph-stationary signals with hidden variables,

    A. Buciulea, S. Rey, and A. G. Marques, “Learning graphs from smooth and graph-stationary signals with hidden variables,” IEEE Transactions on Signal and Information Processing over Networks , vol. 8, pp. 273– 287, 2022

  3. [11]

    Kernel-based structural equation models for topology identification of directed networks,

    Y . Shen, B. Baingana, and G. B. Giannakis, “Kernel-based structural equation models for topology identification of directed networks,” IEEE Transactions on Signal Processing, vol. 65, no. 10, pp. 2503–2516, 2017

  4. [12]

    Graph filters for signal processing and machine learning on graphs,

    E. Isufi, F. Gama, D. I. Shuman, and S. Segarra, “Graph filters for signal processing and machine learning on graphs,” IEEE Transactions on Signal Processing , pp. 1–32, 2024

  5. [13]

    Graph signal processing in the presence of topology uncertainties,

    E. Ceci and S. Barbarossa, “Graph signal processing in the presence of topology uncertainties,” IEEE Transactions on signal processing, vol. 68, pp. 1558–1573, 2020

  6. [14]

    Graph-time convolutional neural networks: Architecture and theoretical analysis,

    M. Sabbaqi and E. Isufi, “Graph-time convolutional neural networks: Architecture and theoretical analysis,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2023

  7. [15]

    Learning stochastic graph neural networks with constrained variance,

    Z. Gao and E. Isufi, “Learning stochastic graph neural networks with constrained variance,” IEEE Transactions on Signal Processing, vol. 71, pp. 358–371, 2023

  8. [16]

    Stochastic graph neural networks,

    Z. Gao, E. Isufi, and A. Ribeiro, “Stochastic graph neural networks,” IEEE Transactions on Signal Processing, vol. 69, pp. 4428–4443, 2021

  9. [17]

    Stability of graph convolutional neural networks to stochastic perturbations,

    ——, “Stability of graph convolutional neural networks to stochastic perturbations,” Signal Processing, vol. 188, p. 108216, 2021

  10. [18]

    Grand: Graph neural diffusion,

    B. Chamberlain, J. Rowbottom, M. I. Gorinova, M. Bronstein, S. Webb, and E. Rossi, “Grand: Graph neural diffusion,” in International Confer- ence on Machine Learning . PMLR, 2021, pp. 1407–1418

  11. [19]

    On the robustness of graph neural diffusion to topology perturbations,

    Y . Song, Q. Kang, S. Wang, K. Zhao, and W. P. Tay, “On the robustness of graph neural diffusion to topology perturbations,” Advances in Neural Information Processing Systems , vol. 35, pp. 6384–6396, 2022

  12. [20]

    Kernels and regularization on graphs,

    A. J. Smola and R. Kondor, “Kernels and regularization on graphs,” in Learning Theory and Kernel Machines: 16th Annual Conference on Learning Theory and 7th Kernel Workshop, COLT/Kernel 2003, 13 Washington, DC, USA, August 24-27, 2003. Proceedings . Springer, 2003, pp. 144–158

  13. [21]

    Kernel-based recon- struction of space-time functions on dynamic graphs,

    D. Romero, V . N. Ioannidis, and G. B. Giannakis, “Kernel-based recon- struction of space-time functions on dynamic graphs,” IEEE Journal of Selected Topics in Signal Processing, vol. 11, no. 6, pp. 856–869, 2017

  14. [22]

    Probabilistic reconstruction of spatio- temporal processes over multi-relational graphs,

    Q. Lu and G. B. Giannakis, “Probabilistic reconstruction of spatio- temporal processes over multi-relational graphs,” IEEE Transactions on Signal and Information Processing over Networks , vol. 7, pp. 166–176, 2021

  15. [23]

    Spatio-temporal inference of dynamical gaussian processes over graphs,

    ——, “Spatio-temporal inference of dynamical gaussian processes over graphs,” in 2021 55th Asilomar Conference on Signals, Systems, and Computers. IEEE, 2021, pp. 1515–1519

  16. [24]

    Gaussian processes on graphs via spectral kernel learning,

    Y .-C. Zhi, Y . C. Ng, and X. Dong, “Gaussian processes on graphs via spectral kernel learning,” IEEE Transactions on Signal and Information Processing over Networks , 2023

  17. [25]

    Non-separable spatio- temporal graph kernels via spdes,

    A. V . Nikitin, S. John, A. Solin, and S. Kaski, “Non-separable spatio- temporal graph kernels via spdes,” in International Conference on Artificial Intelligence and Statistics . PMLR, 2022, pp. 10 640–10 660

  18. [26]

    A survey on graph neural networks for time series: Forecasting, classification, imputation, and anomaly detection,

    M. Jin, H. Y . Koh, Q. Wen, D. Zambon, C. Alippi, G. I. Webb, I. King, and S. Pan, “A survey on graph neural networks for time series: Forecasting, classification, imputation, and anomaly detection,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 46, no....

  19. [27]

    Gated graph recurrent neural networks,

    L. Ruiz, F. Gama, and A. Ribeiro, “Gated graph recurrent neural networks,” IEEE Transactions on Signal Processing , vol. 68, pp. 6303– 6318, 2020

  20. [28]

    Model-Based Deep Learning: On the Intersection of Deep Learning and Optimization,

    N. Shlezinger, Y . C. Eldar, and S. P. Boyd, “Model-Based Deep Learning: On the Intersection of Deep Learning and Optimization,”IEEE Access, vol. 10, pp. 115 384–115 398, 2022

  21. [29]

    Model-Based Deep Learning,

    N. Shlezinger, J. Whang, Y . C. Eldar, and A. G. Dimakis, “Model-Based Deep Learning,” Proceedings of the IEEE, vol. 111, no. 5, pp. 465–499, 2023

  22. [30]

    Robust stochastically- descending unrolled networks,

    S. Hadou, N. NaderiAlizadeh, and A. Ribeiro, “Robust stochastically- descending unrolled networks,” 2023

  23. [31]

    Graph-time trend filtering and unrolling network,

    M. Sabbaqi and E. Isufi, “Graph-time trend filtering and unrolling network,” in 2023 31st European Signal Processing Conference (EU- SIPCO), 2023, pp. 1230–1234

  24. [32]

    Graph unrolling networks: Inter- pretable neural networks for graph signal denoising,

    S. Chen, Y . C. Eldar, and L. Zhao, “Graph unrolling networks: Inter- pretable neural networks for graph signal denoising,” IEEE Transactions on Signal Processing , vol. 69, pp. 3699–3713, 2021

  25. [33]

    Graph signal restoration using nested deep algorithm unrolling,

    M. Nagahama, K. Yamada, Y . Tanaka, S. H. Chan, and Y . C. Eldar, “Graph signal restoration using nested deep algorithm unrolling,” IEEE Transactions on Signal Processing , vol. 70, pp. 3296–3311, 2022

  26. [34]

    Time-varying graph signal inpainting via unrolling networks,

    S. Chen and Y . C. Eldar, “Time-varying graph signal inpainting via unrolling networks,” in ICASSP 2021 - 2021 IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP) , 2021, pp. 8092–8097

  27. [35]

    Unrolling of deep graph total vari- ation for image denoising,

    H. Vu, G. Cheung, and Y . C. Eldar, “Unrolling of deep graph total vari- ation for image denoising,” in ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2021, pp. 2050–2054

  28. [36]

    Observing and tracking bandlimited graph processes from sampled measurements,

    E. Isufi, P. Banelli, P. D. Lorenzo, and G. Leus, “Observing and tracking bandlimited graph processes from sampled measurements,” Signal Processing, vol. 177, p. 107749, 2020

  29. [37]

    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 Transactions on Signal Processing , vol. 70, pp. 1532–1547, 2022

  30. [38]

    Extended Kalman Filter for Graph Signals in Nonlinear Dynamic Systems,

    G. Sagi, N. Shlezinger, and T. Routtenberg, “Extended Kalman Filter for Graph Signals in Nonlinear Dynamic Systems,” in ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2023, pp. 1–5

  31. [39]

    Gsp-kalmannet: Tracking graph signals via neural- aided kalman filtering,

    I. Buchnik, G. Sagi, N. Leinwand, Y . Loya, N. Shlezinger, and T. Routtenberg, “Gsp-kalmannet: Tracking graph signals via neural- aided kalman filtering,” arXiv preprint arXiv:2311.16602 , 2023

  32. [40]

    Kalman Filtering Over Graphs: Theory and Applications,

    L. Shi, “Kalman Filtering Over Graphs: Theory and Applications,” IEEE Transactions on Automatic Control, vol. 54, no. 9, pp. 2230–2234, 2009

  33. [41]

    Unscented Kalman filter of graph signals,

    W. Li, X. Fu, B. Zhang, and Y . Liu, “Unscented Kalman filter of graph signals,” Automatica, vol. 148, p. 110796, 2023

  34. [42]

    Graph-frequency domain kalman filtering for industrial pipe networks subject to measurement outliers,

    L. Su, Z. Han, J. Zhao, and W. Wang, “Graph-frequency domain kalman filtering for industrial pipe networks subject to measurement outliers,” IEEE Transactions on Industrial Informatics , vol. 20, no. 5, pp. 7977– 7985, 2024

  35. [43]

    Stability properties of graph neural networks,

    F. Gama, J. Bruna, and A. Ribeiro, “Stability properties of graph neural networks,” IEEE Transactions on Signal Processing , vol. 68, pp. 5680– 5695, 2020

  36. [44]

    On the transferability of spectral graph filters,

    R. Levie, E. Isufi, and G. Kutyniok, “On the transferability of spectral graph filters,” in 2019 13th International conference on Sampling Theory and Applications (SampTA) . IEEE, 2019, pp. 1–5

  37. [45]

    Capp ´e, E

    O. Capp ´e, E. Moulines, and T. Ryd ´en, Inference in Hidden Markov Models, ser. Springer Series in Statistics. Springer New York, NY , 2005

  38. [46]

    Maximum likelihood from incomplete data via the em algorithm,

    A. P. Dempster, N. M. Laird, and D. B. Rubin, “Maximum likelihood from incomplete data via the em algorithm,” Journal of the royal statistical society: series B (methodological) , vol. 39, no. 1, pp. 1–22, 1977

  39. [47]

    Statistical inference for probabilistic functions of finite state markov chains,

    L. E. Baum and T. Petrie, “Statistical inference for probabilistic functions of finite state markov chains,” The annals of mathematical statistics, vol. 37, no. 6, pp. 1554–1563, 1966

  40. [48]

    Data-driven initialization of deep learning solvers for hamilton-jacobi-bellman pdes,

    A. Borovykh, D. Kalise, A. Laignelet, and P. Parpas, “Data-driven initialization of deep learning solvers for hamilton-jacobi-bellman pdes,” IFAC-PapersOnLine, vol. 55, no. 30, pp. 168–173, 2022, 25th Interna- tional Symposium on Mathematical Theory of Networks and Systems MTNS 2022

  41. [49]

    New Results in Linear Filtering and Prediction Theory,

    R. E. Kalman and R. S. Bucy, “New Results in Linear Filtering and Prediction Theory,” Journal of Basic Engineering , vol. 83, no. 1, pp. 95–108, 1961

  42. [50]

    Graphs, convolutions, and neural networks: From graph filters to graph neural networks,

    F. Gama, E. Isufi, G. Leus, and A. Ribeiro, “Graphs, convolutions, and neural networks: From graph filters to graph neural networks,” IEEE Signal Processing Magazine , vol. 37, no. 6, pp. 128–138, 2020

  43. [51]

    R. B. Bapat, Graphs and matrices . Springer, 2010, vol. 27

  44. [52]

    Diffusion convolutional re- current neural network: Data-driven traffic forecasting,

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

  45. [53]

    G. E. Box, G. M. Jenkins, G. C. Reinsel, and G. M. Ljung, Time series analysis: forecasting and control . John Wiley & Sons, 2015

  46. [54]

    Forecasting time series with varma recursions on graphs,

    E. Isufi, A. Loukas, N. Perraudin, and G. Leus, “Forecasting time series with varma recursions on graphs,” IEEE Transactions on Signal Processing, vol. 67, no. 18, pp. 4870–4885, 2019

  47. [55]

    Graph wavenet for deep spatial-temporal graph modeling,

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

  48. [56]

    Spatial temporal graph convolutional networks for skeleton-based action recognition,

    S. Yan, Y . Xiong, and D. Lin, “Spatial temporal graph convolutional networks for skeleton-based action recognition,” Proceedings of the AAAI Conference on Artificial Intelligence , vol. 32, no. 1, Apr. 2018

  49. [57]

    Deep reinforcement learning for the real time control of stormwater systems,

    A. Mullapudi, M. J. Lewis, C. L. Gruden, and B. Kerkez, “Deep reinforcement learning for the real time control of stormwater systems,” Advances in Water Resources , vol. 140, p. 103600, 2020

  50. [58]

    Transferable and data efficient metamodeling of storm water system nodal depths using auto-regressive graph neural networks,

    A. Garz ´on, Z. Kapelan, J. Langeveld, and R. Taormina, “Transferable and data efficient metamodeling of storm water system nodal depths using auto-regressive graph neural networks,” Water Research, vol. 266, p. 122396, 2024

  51. [59]

    Making waves: Towards data-centric water engineering,

    G. Fu, D. Savic, and D. Butler, “Making waves: Towards data-centric water engineering,” Water Research, vol. 256, p. 121585, 2024

  52. [60]

    Towards transfer- able metamodels for water distribution systems with edge-based graph neural networks,

    B. Kerimov, R. Taormina, and F. Tscheikner-Gratl, “Towards transfer- able metamodels for water distribution systems with edge-based graph neural networks,” Water Research, vol. 261, p. 121933, 2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.