Pith. sign in

REVIEW 4 major objections 5 minor 39 references

Multilevel Graph Wavelet Compressed Sensing with Scale-Aware Neural Recovery

T0 review · 4 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read Graph Wavelet Compressed Sensing reconstructs mesh signals from 5% of their wavelet coefficients, beating classical graph-sampling baselines on spectrally structured PDE data.

desk verdict A genuinely new combination of wavelet-domain sampling and GNN recovery with one strong win and several losses, but the missing parameter-matched baseline undercuts the headline claim. read the letter →

arxiv 2607.20857 v1 pith:KRZHEJLC submitted 2026-07-23 cs.LG cs.AI

Amirhossein Nouranizadeh , Sarang Rajendra Patil , Alan John Varghese , Varsha Narayanan , Amit Chakraborty , Mengjia Xu This is my paper
classification cs.LGcs.AI
keywords graphwavelettransformcompressedsensingsignalsamplingneuralnetworksscientificmachinelearningmeshdatacompressionmultilevelimportancescale-awarerecovery
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 proposes Graph Wavelet Compressed Sensing (GWCS), a framework that compresses graph-structured scientific data—PDE simulation fields on unstructured meshes—by transforming each signal into a sparse spectral graph wavelet representation, retaining only a small fraction of the largest-energy wavelet coefficients via a nonparametric multilevel importance sampler (MLIS), and reconstructing the full signal with a scale-aware graph neural network (NIGWT). The central claim is that a fixed wavelet prior plus learned recovery achieves high reconstruction fidelity at a 5% coefficient budget, outperforming classical graph signal sampling methods and a learned sparse autoencoder on data whose energy concentrates in a few scales. The paper demonstrates this on synthetic approximately band-limited signals and four PDE datasets, with the clearest gains on viscoelastic instabilities and approximate band-limited graphs. A sympathetic reader would care because offline compression of simulation data is a bottleneck in scientific machine learning, and GWCS offers an interpretable, training-stable alternative to learned latent representations.

What carries the argument

The Spectral Graph Wavelet Transform (SGWT)—a filterbank of one low-pass and four band-pass spectral filters approximated by Chebyshev polynomials—provides the sparse, multiscale representation. The Multilevel Importance Sampling (MLIS) module splits the coefficient budget across scales by per-scale energy and then samples within scales with probability proportional to squared coefficient magnitude, producing a binary mask. The Neural Inverse Graph Wavelet Transform (NIGWT) is an Encode-Process-Decode GNN that takes the masked coefficients and learnable scale embeddings and adds a residual connection to the pseudo-inverse SGWT, so the network only learns the correction beyond the linear inve

What would settle it

An oracle experiment would settle the central claim: for the same budget, compute the best possible coefficient selection via a greedy backward elimination or exhaustive search on a few test signals, and compare its reconstruction error to MLIS. If the oracle is substantially better on spectrally structured data, MLIS is not near-optimal and GWCS's reported gains come from the heuristic, not the wavelet prior. Conversely, on a signal whose wavelet energy is evenly spread across all scales, if GWCS at c=5% cannot beat a simple global top-m-absolute-coefficient reconstruction, the within-scale e

Watch

Extended reading notes

Core claim

GWCS's central discovery is that a fixed, energy-based multilevel sampling of spectral graph wavelet coefficients, combined with a neural network that learns a scale-aware inverse transform, can reconstruct graph signals from as few as 1–5% of their wavelet coefficients. The MLIS sampler allocates the retention budget across wavelet scales in proportion to per-scale energy and, within each scale, keeps coefficients with the largest squared magnitude; the NIGWT network then refines the pseudo-inverse wavelet reconstruction with a learnable residual. On synthetic approximately band-limited signals, GWCS beats all classical graph-signal-sampling baselines, which are given oracle bandwidth knowl

Load-bearing premise

The MLIS heuristic—that keeping the largest-magnitude coefficients within each scale, with the budget split in proportion to per-scale energy, is a near-optimal way to minimize reconstruction error under a fixed budget—is the load-bearing premise; if it fails, GWCS's advantage over learned latents disappears.

Editorial extensions

If this is right

  • On spectrally concentrated fields, GWCS reaches RMSE below 0.002 at a 5% coefficient budget, about 2x lower than the best learned baseline and 3x lower than classical sampling.
  • A trained GWCS model operates inductively: new unseen signals on new graphs are compressed and reconstructed with a single forward pass, no per-signal optimization.
  • The retained wavelet coefficients form an interpretable compressed representation, showing which scales and nodes carry the energy, which may be useful for downstream scientific analysis.
  • The comparison with a learned sparse autoencoder shows that the wavelet prior helps most when energy is concentrated in few scales; on broadband data the learned latent remains superior, clarifying when each approach is preferable.
  • All classical graph signal sampling methods, even with oracle bandwidth knowledge, are outperformed by both GWCS and the learned autoencoder on the PDE testbeds, supporting learned recovery over classic sampling in this regime.

Reading between the lines

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

  • The MLIS heuristic could be replaced by a learned sampler that allocates budget across scales based on predicted reconstruction gain; this might close the gap on broadband datasets like the turbulent radiative layer, where the fixed energy-proportional split currently loses to the learned autoencoder.
  • Extending MLIS and NIGWT to include temporal dimensions, as the authors list as future work, could yield a spatiotemporal compressor that exploits both spatial and temporal sparsity, likely improving compression ratios for time-series simulation data.
  • The residual IGWT connection is a general design pattern: any linear inverse transform with known error can be bootstrapped by a small GNN, potentially applicable outside graph wavelets (e.g., in tomography or other compressed sensing tasks).
  • A testable design suggestion is to increase the number of wavelet scales or adapt the kernels to the dataset's spectrum; if the fixed five-filter bank is the main constraint on the turbulent cases, a richer filterbank should improve GWCS there.
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

4 major / 5 minor

Summary. GWCS is a learning-based offline compression framework for graph signals. It transforms the signal via the spectral graph wavelet transform (Chebyshev-approximated), then uses MLIS to retain a fixed budget of coefficients: the budget is split across scales in proportion to per-scale energy (Eq. 4.2), and within each scale coefficients are kept with probability proportional to squared magnitude (Eq. 4.3). NIGWT reconstructs the signal from the sparse coefficients via a scale-aware Encode-Process-Decode GNN with a residual IGWT connection. A wavelet-free Sparse Graph Autoencoder (SGAE) is introduced as a baseline. Experiments on synthetic approximately band-limited signals over random graphs and on four PDE datasets (Turbulent Radiative Layer, Viscoelastic Instability, Kolmogorov Flow, Dynamic Stall) compare GWCS with classical graph-signal-sampling methods, GXN, and SGAE. The paper claims that GWCS achieves high reconstruction fidelity and substantial data compression, with the strongest quantitative result on Viscoelastic Instability (RMSE 0.0016 vs. 0.0035 for SGAE-50K at c=5%) and competitive performance on Kolmogorov Flow, while SGAE leads on Turbulent and Dynamic Stall.

Significance. The paper proposes a practical, interpretable way to compress graph-structured PDE data using multiscale wavelet sparsity plus a small neural corrector. Strengths include a simple nonparametric sampling rule, a clearly described network architecture, open code, and the use of standard PDE benchmarks. If the Viscoelastic advantage survives a controlled parameter-matched comparison, the framework is a useful offline compression tool for spectrally structured fields. However, the current evidence is insufficient: the promised lightweight GWCS variant is never reported, so the headline gain over SGAE may be capacity rather than the wavelet prior; and the abstract overstates scope relative to the paper's own Dynamic Stall and Turbulent results. The paper's body discussion is more balanced than its abstract.

major comments (4)
  1. [§5.2 and Tables 4.1/4.2] Section 5.2 promises a lightweight GWCS variant (D=96, L=3, ~99K parameters) for a parameter-matched comparison against SGAE-99K, but no reported table or appendix contains results for this variant; all reported GWCS rows appear to be the 993K model. The headline Viscoelastic Instability comparison (Table 4.2: GWCS 0.0016 vs. SGAE-50K 0.0035) therefore confounds the wavelet-domain sampling and the 10–20x parameter difference. Because the paper's central claim is that the wavelet prior drives the advantage, the promised controlled comparison is necessary; without it the claim is unsupported.
  2. [Abstract and Table 4.2] The abstract claims GWCS 'achieves high reconstruction fidelity and substantial data compression compared to existing benchmarks' without qualification. The paper's own data contradict this blanket statement: on Dynamic Stall, GWCS RMSE is 9119±1700, worse than BSGDA (5368±1040) and RSBS (6092±1203); on Turbulent Radiative Layer, GWCS (7.06) is worse than SGAE-99K (4.78); on Kolmogorov Flow, GWCS (0.316) is worse than SGAE-50K (0.287). The body contains a more nuanced and accurate discussion, but the abstract should be revised to state the conditions under which GWCS is best (e.g., spectrally concentrated energy).
  3. [Table 4.2 footnote and §5.3] Classical GSS baselines are evaluated on only 50 test samples, while neural methods use the full test set (Table 4.2 footnote). This makes the mean RMSE comparisons across methods not fully comparable, especially for datasets with high variability. On Dynamic Stall, for example, the reported standard deviations are large relative to differences. Please report results on a common test subset (or report neural-model metrics on the same 50-sample subset) so that the comparisons are apples-to-apples.
  4. [§4.1 MLIS heuristic] The MLIS energy-based rule is the core of the method, but the manuscript only provides an intuitive justification (Eq. 4.3 paragraph) and acknowledges in §5.6 that it is a heuristic. The paper's failure cases (Turbulent Radiative Layer, Dynamic Stall) could be due to the fixed 5-filter SGWT, or to the inadequacy of magnitude-based coefficient retention. A simple ablation comparing MLIS to random retention within the same per-scale budget, or to an oracle that selects coefficients to minimize reconstruction error, would directly test the premise and strengthen the central claim.
minor comments (5)
  1. [References] References [3] and [4] are identical (both cite Anis, Gadde, and Ortega, 'Efficient sampling set selection for bandlimited graph signals using graph spectral proxies').
  2. [Table 4.1] The column structure is confusing: the top row gives compression ratios 5%, 25%, 50% but the sub-row gives '1% / 5% / 10% coefficients'. Clarify that the coefficient budget is c times the number of nodes (i.e., 1% of coefficients at c=5%? Or clarify the relationship); currently it is easy to misread.
  3. [Appendix B and Figure E.3] Appendix B says N is sampled uniformly from [100,625], but Figure E.3 reports N=65 for a representative Erdős–Rényi graph. Check the range or the figure caption.
  4. [Appendix E] Figure E.5 caption says 'test sample 47' in the text body, then lists 'sample 48' in subcaptions (a)–(f). Standardize sample indices to avoid ambiguity, especially because Figure E.5 in the main text references sample 48.
  5. [§5.4] The sentence 'GWCS outperforms all GSS baselines ... without oracle bandwidth knowledge' could be clearer: the GSS baselines are given oracle bandwidth (Appendix C), while GWCS does not require it. State this explicitly in the main text.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: neither MLIS nor NIGWT reduces to a fitted input, and the sole self-citation is non-load-bearing.

full rationale

I walked the claimed derivation chain: MLIS (Eqs. 4.1-4.3) is an adaptive sampling heuristic defined directly from the signal's own wavelet coefficients W = Ψx, explicitly justified by the intuition that 'dropping a large-magnitude coefficient increases the reconstruction error more than dropping a near-zero one.' It is not a fitted parameter later reported as a prediction, and the paper does not claim it is optimal. NIGWT (Eqs. 4.5-4.9) is a learned residual reconstruction: x_hat = IGWT(W_tilde) + fDec(Z^(L)), trained by the MSE objective (4.11). This is standard supervised regression from sparse coefficients to the original signal, with no quantity fitted to a subset and then used to 'predict' a closely related quantity by construction. The only author-overlap reference is [27], cited in Section 2.3 as one of three general citations for 'multi-scale architectures that capture hierarchical dependencies [40,17,27]'; it is not used to justify any design choice, uniqueness claim, or ansatz, so it is non-load-bearing. I assign score 1 only to mark this peripheral self-citation, not because any derivation is circular. Separately, the paper promises a parameter-matched comparison that never appears: Section 5.2 states 'A lightweight variant (D=96, L=3, ≈99,000 parameters) is used for parameter-matched comparison against SGAE-99K,' yet Tables 4.1, 4.2, D.1, and D.2 report no lightweight GWCS results, and Table 4.2's classical-GSS values are computed on only 50 test samples while neural methods use the full test set. Also, on Dynamic Stall GWCS (9119±1700) is worse than SGAE-99K (2480±744) and BSGDA (5368±1040), contradicting the abstract's blanket comparative claim. These are correctness, fairness, and omitted-result concerns, not circularity: they do not make any reported number equivalent to its input by construction. The framework is evaluated against external benchmarks (WELL, PBFM, THGSP GSS baselines, and independently trained SGAE/GXN baselines), so the central comparison is not driven by a self-citation chain.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

GWCS rests on standard spectral graph theory plus two paper-specific choices: the energy-magnitude sampling heuristic and a fixed 5-filter Chebyshev SGWT. The free parameters are design choices, hand-selected rather than fitted to optimize reported metrics, but they affect the outcome. No constants are fitted to data in a derivation, and no new physical entities are introduced.

free parameters (4)
  • compression ratio c = 0.05 (5%)
    Hand-selected budget for all main experiments; controls how many wavelet coefficients or latent entries are retained.
  • number of SGWT scales J = 4 band-pass plus 1 low-pass (F=5)
    Fixed filterbank; not tuned per dataset, but the number of scales directly determines how much broadband energy is discarded.
  • Chebyshev polynomial degree K = 100
    Chosen for approximating the SGWT filters; affects accuracy and computational cost of the forward/inverse transforms.
  • architecture hyperparameters = D=256, L=5 for GWCS; D_SGAE=128, L=4 for SGAE
    Model capacities selected by hand; SGAE-99K vs SGAE-50K show that capacity changes ranking on some datasets.
assumptions (5)
  • standard math SGWT filterbank satisfies admissibility and the pseudo-inverse Psi^dagger exists and is computable via Chebyshev approximation.
    Invoked in Sections 3.2, 3.3, and 4.2 to justify forward and inverse graph wavelet transforms.
  • domain assumption Physical PDE fields on meshes are approximately sparse in the fixed five-scale graph wavelet domain.
    Motivates the whole compression approach; cited via [36, 32] and confirmed only for some datasets in Section 5.5.
  • ad hoc to paper High-magnitude wavelet coefficients are the right coefficients to keep at a fixed budget, and per-scale energy is the right way to allocate the budget.
    MLIS heuristic, Section 4.1, Eqs. (4.1)-(4.3); stated as intuitive, not proven, and contradicted on broadband datasets.
  • domain assumption Chebyshev degree K=100 gives a sufficiently accurate approximation of the forward and inverse SGWT, including the pseudo-inverse.
    Used throughout the experiments; no error analysis is provided for K=100 on the specific graph sizes and Laplacian spectra.
  • domain assumption A model trained on snapshots from one mesh/distribution generalizes to test snapshots from the same mesh/distribution.
    Required for the inductive evaluation on the four PDE datasets; not tested across different meshes or graph sizes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multilevel Graph Wavelet Compressed Sensing with Scale-Aware Neural Recovery." pith.science (2026). https://pith.science/paper/KRZHEJLC

@misc{pith2026260720857,
  author       = {Pith},
  title        = {Pith review of: Multilevel Graph Wavelet Compressed Sensing with Scale-Aware Neural Recovery},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KRZHEJLC}},
  note         = {Machine review of arXiv:2607.20857}
}
read the original abstract

Scientific machine learning methods such as neural operators and physics-informed neural networks have advanced engineering applications and inverse problems, but their training typically requires large volumes of simulated data. This makes data preparation and model training expensive. We propose Graph Wavelet Compressed Sensing (GWCS), a learning-based framework for offline compression of graph signals by representing them as sparse, interpretable wavelet-domain representations using the spectral graph wavelet transform. The framework combines a nonparametric multilevel importance sampler, which retains high-energy wavelet coefficients within each scale for a given compression ratio, with a scale-aware graph neural network that reconstructs the signal from the sparse coefficients. We evaluate the proposed framework on synthetic approximately band-limited graph signals over random graphs and four PDE simulation datasets over meshes, which include Turbulent Radiative Layer, Viscoelastic Instability, Kolmogorov Flow, and Dynamic Stall. We compare against graph signal sampling methods and graph autoencoder baselines. Results demonstrate that the framework achieves high reconstruction fidelity and substantial data compression compared to existing benchmarks.

Figures

Figures reproduced from arXiv: 2607.20857 by the authors.

Figure 4.1
Figure 4.1. Overview of the GWCS Framework. 1) The input graph signal [PITH_FULL_IMAGE:figures/full_fig_p004_4_1.png] view at source ↗
Figure 4.2
Figure 4.2. Overview of the SGAE Framework. The raw signal x ∈ R N is encoded directly by the Sparse Encoder into a sparse latent Z˜ ∈ R N×DSGAE (DSGAE = 128; cN entries active). The Reconstruction Model (Graph Decoder) maps Z˜ back to the signal domain. Training Objectives. Both GWCS and SGAE are trained end-to-end by minimizing the same MSE re￾construction loss over the dataset with the Stochastic Gradient Descent (SGD). For … view at source ↗
Figure 5
Figure 5. (b), all three neural methods cluster well [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figures from the paper (2 more)
Figure 5.1
Figure 5.1. Figure 5.1: Qualitative reconstruction of the Turbulent Radiative Layer density field (test sample 48, compression [PITH_FULL_IMAGE:figures/full_fig_p008_5_1.png]
Figure 5.2
Figure 5.2. Figure 5.2: Per-sample RMSE trajectories (physical units). [PITH_FULL_IMAGE:figures/full_fig_p008_5_2.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

39 extracted references · 5 linked inside Pith

  1. [1]

    Adcock, A

    B. Adcock, A. C. Hansen, C. Poon, and B. Roman,Breaking the coherence barrier: A new theory for compressed sensing, in Forum of mathematics, sigma, vol. 5, Cambridge University Press, 2017, p. e4

    2017

  2. [2]

    Adcock, A

    B. Adcock, A. C. Hansen, and B. Roman, The quest for optimal sampling: Computationally efficient, structure-exploiting measurements for com- pressed sensing, in Compressed Sensing and its Ap- plications: MATHEON Workshop 2013, Springer, 2015, pp. 143–167

    2013

  3. [4]

    A. Anis, A. Gadde, and A. Ortega,Efficient sampling set selection for bandlimited graph signals using graph spectral proxies, IEEE Transactions on Signal Processing, 64 (2016), pp. 3775–3789

    2016

  4. [5]

    Y. Bai, F. W ang, G. Cheung, Y. Nakatsukasa, and W. Gao,Fast graph sampling set selection using gershgorin disc alignment, IEEE Transactions on signal processing, 68 (2020), pp. 2419–2434

    2020

  5. [6]

    Baldan, Q

    G. Baldan, Q. Liu, A. Guardone, and N. Thuerey,Physics vs distributions: Pareto opti- mal flow matching with physics constraints, in The Fourteenth International Conference on Learning Representations, 2026

    2026

  6. [7]

    Barabási and R

    A.-L. Barabási and R. Albert,Emergence of scaling in random networks, science, 286 (1999), pp. 509–512

    1999

  7. [8]

    P. W. Battaglia, J. B. Hamrick, V. Bapst, A. Sanchez-Gonzalez, V. Zambaldi, M. Ma- linowski, A. Tacchetti, D. Raposo, A. San- toro, R. F aulkner, et al.,Relational inductive biases, deep learning, and graph networks, arXiv preprint arXiv:1806.01261, (2018)

    arXiv 2018

  8. [9]

    E. J. Candès, J. Romberg, and T. Tao,Robust uncertainty principles: Exact signal reconstruction from highly incomplete frequency information, IEEE Transactions on information theory, 52 (2006), pp. 489–509

    2006

Show all 39 references
  1. [10]

    G.-H. Chen, J. Tang, and S. Leng,Prior image constrained compressed sensing (piccs): a method to accurately reconstruct dynamic ct images from highly undersampled projection data sets, Medical physics, 35 (2008), pp. 660–663

    2008

  2. [11]

    S. Chen, Y. C. Eldar, and L. Zhao,Graph unrolling networks: Interpretable neural networks for graph signal denoising, IEEE Transactions on Signal Processing, 69 (2021), pp. 3699–3713

    2021

  3. [12]

    S. Chen, M. Li, and Y. Zhang,Sampling and recovery of graph signals based on graph neural networks, arXiv preprint arXiv:2011.01412, (2020)

    2011 arXiv

  4. [13]

    S. Chen, R. V arma, A. Sandryhaila, and J. Kov ačević,Discrete signal processing on graphs: Sampling theory<? pub _newline=""?, IEEE trans- actions on signal processing, 63 (2015), pp. 6510– 6523

    2015

  5. [14]

    Deng,Thgsp: A pytorch-based graph signal processing library

    B. Deng,Thgsp: A pytorch-based graph signal processing library. https://github.com/bwdeng20/ thgsp, 2021

    2021

  6. [15]

    D. L. Donoho,Compressed sensing, IEEE Trans- actions on information theory, 52 (2006), pp. 1289– 1306

    2006

  7. [16]

    ERDdS and A

    P. ERDdS and A. R&wi,On random graphs i, Publ. math. debrecen, 6 (1959), p. 18

    1959

  8. [17]

    Gao and S

    H. Gao and S. Ji,Graph u-nets, in international conference on machine learning, PMLR, 2019, pp. 2083–2092

    2019

  9. [18]

    Hamilton, Z

    W. Hamilton, Z. Ying, and J. Leskovec, Inductive representation learning on large graphs, Advances in neural information processing systems, 30 (2017)

    2017

  10. [19]

    D. K. Hammond, P. V andergheynst, and R. Gribonv al,Wavelets on graphs via spectral graph theory, Applied and Computational Harmonic Analysis, 30 (2011), pp. 129–150

    2011

  11. [20]

    T. N. Kipf and M. Welling,Variational graph auto-encoders, arXiv preprint arXiv:1611.07308, (2016)

    2016 arXiv

  12. [21]

    T. N. Kipf and M. Welling,Semi-supervised classification with graph convolutional networks, in International Conference on Learning Representa- tions (ICLR), 2017

    2017

  13. [22]

    Kov achki, Z

    N. Kov achki, Z. Li, B. Liu, K. Azizzade- nesheli, K. Bhattacharya, A. Stuart, and A. Anandkumar,Neural operator: Learning maps between function spaces with applications to pdes, Journal of Machine Learning Research, 24 (2023), pp. 1–97

    2023

  14. [23]

    Leary, Z

    R. Leary, Z. Saghi, P. A. Midgley, and D. J. Holland,Compressed sensing electron tomography, Ultramicroscopy, 131 (2013), pp. 70–91

    2013

  15. [24]

    L. Lu, P. Jin, G. Pang, Z. Zhang, and G. E. Karniadakis,Learning nonlinear operators via deeponet based on the universal approximation theorem of operators, Nature machine intelligence, 3 (2021), pp. 218–229

    2021

  16. [25]

    Lustig, D

    M. Lustig, D. Donoho, and J. M. Pauly, Sparse mri: The application of compressed sensing for rapid mr imaging, Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine, 58 (2007), pp. 1182–1195

    2007

  17. [26]

    A. G. Marques, S. Segarra, G. Leus, and A. Ribeiro,Sampling of graph signals with succes- sive local aggregations, IEEE Transactions on Signal Processing, 64 (2015), pp. 1832–1843

    2015

  18. [27]

    Nouranizadeh, M

    A. Nouranizadeh, M. Matinkia, M. Rahmati, and R. Saf abakhsh,Maximum entropy weighted independent set pooling for graph neural networks, arXiv preprint arXiv:2107.01410, (2021)

    2021 arXiv

  19. [28]

    Ohana, M

    R. Ohana, M. McCabe, L. Meyer, R. Morel, F. Agocs, M. Beneitez, M. Berger, B. Burkhart, S. Dalziel, D. Fielding, et al., The well: a large-scale collection of diverse physics simulations for machine learning, Advances in Neu- ral Information Processing Systems, 37 (2024), pp. ...

    2024

  20. [29]

    Ortega,Introduction to graph signal processing, Cambridge University Press, 2022

    A. Ortega,Introduction to graph signal processing, Cambridge University Press, 2022

    2022

  21. [30]

    G. Puy, N. Tremblay, R. Gribonv al, and P. V andergheynst,Random sampling of bandlim- ited signals on graphs, Applied and Computational Harmonic Analysis, 44 (2018), pp. 446–475

    2018

  22. [31]

    Raissi, P

    M. Raissi, P. Perdikaris, and G. E. Kar- niadakis,Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differ- ential equations, Journal of Computational physics, 378 (2019), pp. 686–707

    2019

  23. [32]

    Ricaud, D

    B. Ricaud, D. I. Shuman, and P. V an- dergheynst,On the sparsity of wavelet coefficients for signals on graphs, in Wavelets and Sparsity XV, vol. 8858, SPIE, 2013, pp. 422–428

    2013

  24. [33]

    Roman, A

    B. Roman, A. Hansen, and B. Adcock,On asymptotic structure in compressed sensing, arXiv preprint arXiv:1406.4178, (2014)

    2014 arXiv

  25. [34]

    Sakiyama, Y

    A. Sakiyama, Y. Tanaka, T. Tanaka, and A. Ortega,Eigendecomposition-free sampling set selection for graph signals, IEEE Transactions on Signal Processing, 67 (2019), pp. 2679–2692

    2019

  26. [35]

    Sanchez-Gonzalez, J

    A. Sanchez-Gonzalez, J. Godwin, T. Pf aff, R. Ying, J. Leskovec, and P. Battaglia, Learning to simulate complex physics with graph networks, in International conference on machine learning, PMLR, 2020, pp. 8459–8468

    2020

  27. [36]

    Schneider and O

    K. Schneider and O. V. V asilyev,Wavelet methods in computational fluid dynamics, Annual review of fluid mechanics, 42 (2010), pp. 473–503

    2010

  28. [37]

    D. I. Shuman, S. K. Narang, P. Frossard, A. Ortega, and P. V andergheynst,The emerg- ing field of signal processing on graphs: Extend- ing high-dimensional data analysis to networks and other irregular domains, IEEE signal processing magazine, 30 (2013), pp. 83–98

    2013

  29. [38]

    Tremblay, P.-O

    N. Tremblay, P.-O. Amblard, and S. Barthelmé,Graph sampling with deter- minantal processes, in 2017 25th European signal processing conference (EUSIPCO), IEEE, 2017, pp. 1674–1678

    2017

  30. [39]

    V asw ani, N

    A. V asw ani, N. Shazeer, N. Parmar, J. Uszko- reit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin,Attention is all you need, Advances in neural information processing systems, 30 (2017)

    2017

  31. [40]

    Z. Ying, J. You, C. Morris, X. Ren, W. Hamil- ton, and J. Leskovec,Hierarchical graph repre- sentation learning with differentiable pooling, Ad- vances in neural information processing systems, 31 (2018). A Chebyshev polynomial approximation of SGWT.Concretely, we map the spec...

    2018

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.