Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

Point Cloud Denoising With Fine-Granularity Dynamic Graph Convolutional Networks

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper claims that denoising 3-D point clouds should be treated as a continuous dynamical system, and that micro-step temporal graph convolution, a learned Riemannian metric, and stable Bernstein spectral filters let one network beat…

desk verdict Solid empirical architecture paper whose Riemannian-metric story doesn't survive contact with the equations, but whose denoising results still deserve review. read the letter →

arxiv 2411.14158 v1 pith:QWMTCP3L submitted 2024-11-21 cs.CV

Wenqiang Xu , Wenrui Dai , Duoduo Xue , Ziyang Zheng , Chenglin Li , Junni Zou , Hongkai Xiong This is my paper · ORCID
classification cs.CV
keywords pointclouddenoisinggraphconvolutionalnetworkmicro-steptemporalconvolutionneuralODERiemannianmetricBernsteinpolynomialspectralfilteringdynamic
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 tries to establish that point-cloud denoising is best modeled as a continuous-time dynamic process, not as a stack of discrete graph convolutions. It proposes GD-GCN, which splits each message-passing step into many micro steps governed by a neural ODE, builds its graph with a learned approximate Riemannian metric instead of plain Euclidean distance, and filters in the graph spectral domain with Bernstein polynomials whose coefficients are constrained so the system stays bounded. The payoff, if correct, is that one network beats ten state-of-the-art denoisers on most standard noise levels in both supervised and unsupervised modes, with the largest reported gain at 1% noise on sparse clouds, and generalizes to real LiDAR data. That matters because denoising is a prerequisite for surface reconstruction, rendering, and downstream perception, and the method claims all this without per-noise-level retuning.

What carries the argument

The load-bearing object is MST-GConv, a residual graph convolution written as a dynamical system: $Z^{l} = Z^{l-1} + \delta((A_G^l - I)Z^{l-1}\Theta^l)\tau^l$ with micro-step size $\tau^l \ll 1$, solved numerically as a neural ODE or neural PDE (Eqs. 3–5). Around it are two supporting mechanisms: (1) geometric graph construction via the approximate Riemannian metric $G = I + \alpha^2(\nabla z(p))^\top(\nabla z(p))$ (Eq. 7), where the Jacobian is replaced by a multi-head attention output $J_\theta$ (Eq. 8) and edge weights are $\tilde{a}_{ij} = \exp(-|\ell_{ij}|^2/(2\delta^2))$; and (2) a graph spectral filter built on the Bernstein polynomial basis $B_K(\lambda) = \sum_{k=0}^K \theta_k \binom{K}{k}(1-\lambda)^{K-k}\lambda^k$ with coefficient constraints giving $0 < B_K(\lambda) \le 1$ (Proposition 1, Eq. 9), combined with symmetric channel mixing matrices $W_1^t, W_2^t$ that control per-channel frequency scaling and shifting (Eqs. 10–11). The machinery's role is to let the network capture topology changes during denoising, separate geometric regions, and keep the evolution numerically stable while avoiding low-frequency dominance.

What would settle it

Train GD-GCN, then on a held-out noisy point cloud compute the finite-difference Jacobian of the map from input 3-D coordinates to the $l$-th layer feature $z(p)$ for a set of points, and compare its action on neighbor displacement vectors with the attention-based $J_\theta$ used in Eq. (8); a systematic mismatch (wrong dimension, sign flip, or large directional error) would falsify the claim that the graph construction is geometric.

Watch

Extended reading notes

Core claim

The paper's central claim is that point cloud denoising should be treated as a continuous-time dynamical system rather than as a stack of discrete graph convolution layers. It introduces micro-step temporal graph convolution (MST-GConv), which replaces the one-hop message passing of a residual GCN with many small updates governed by a neural ODE, so that noisy points evolve smoothly toward the underlying surface. To build the graph for these updates, the method learns an approximate Riemannian metric $G = I + \alpha^2(\nabla z(p))^\top(\nabla z(p))$ and measures edge weights by the induced distance, separating points whose Euclidean proximity is deceptive (e.g., across sharp edges). To keep the evolution stable, it applies a Bernstein-polynomial spectral filter $B_K(\lambda)$ with coefficients constrained so that $0 < B_K(\lambda) \le 1$, giving bounded-input bounded-output stability, plus symmetric channel mixing matrices that scale and shift frequency components to avoid low-frequency dominance. On supervised and unsupervised denoising benchmarks, the paper reports lower Chamfer distance and Earth Mover's distance than ten baselines across most noise levels, for example Chamfer distance $4.39 \times 10^{-5}$ versus the previous best $6.02 \times 10^{-5}$ on sparse 10K point clouds with 1% Gaussian noise.

Load-bearing premise

The whole 'Riemannian metric' graph construction assumes that the multi-head attention output $J_\theta$ in Eq. (8) is a valid approximation to the true Jacobian $\nabla z(p)$ of the learned feature map, with no error bound or dimension check; if it is not, the graph is not actually measuring Riemannian distance.

Editorial extensions

If this is right

  • On the 10K sparse test set at 1% Gaussian noise, GD-GCN reports Chamfer distance $4.39 \times 10^{-5}$ versus $6.02 \times 10^{-5}$ for the best prior method, and it leads on Earth Mover's distance as well.
  • At termination times $0.25T$, $0.5T$, $0.75T$, and $T$, the extracted intermediate states show the denoised cloud progressively approaching the ground-truth surface, confirming the continuous-evolution interpretation.
  • The Bernstein-polynomial filter bounds the filtered eigenvalues to $(0,1]$, so the dynamic system is BIBO stable and cannot diverge or amplify noise during long integration.
  • On real Paris-rue-Madame LiDAR data, the method achieves lower CD, HD, EMD, and RMSD than the compared unsupervised models, indicating that the learned dynamics carry over to realistic acquisition noise.

Reading between the lines

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

  • The same micro-step temporal graph convolution could be transplanted to other graph-based geometry tasks (segmentation, upsampling, normal estimation), since the continuous-time formulation is task-agnostic; a testable extension would be replacing the point-coordinate signal with other features.
  • If the attention-based Jacobian truly approximates the Riemannian metric, the geometric graph construction should improve any distance-based point-cloud method, not just this architecture; an ablation that swaps the learned metric for a frozen Euclidean metric on the same backbone would isolate that contribution further.
  • The stability bound suggests the network can be integrated for longer times or with more micro-steps than tested, which the paper does not push to its limit; increasing $T$ beyond 1 at high noise levels would reveal whether the bound translates to better asymptotic fitting.
  • The paper's own results show the unsupervised version loses to the supervised one at 3% noise on sparse clouds; this hints that the learned metric and spectral filter may need noise-level-aware adaptation, a direction the authors do not explore.
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 / 6 minor

Summary. The paper proposes GD-GCN, a graph-convolutional network for 3-D point cloud denoising. The method has three advertised components: (i) MST-GConv, a micro-step temporal graph convolution interpreted through the lens of neural ODEs/PDEs; (ii) a geometric graph construction that claims to approximate a Riemannian metric from an attention-based Jacobian estimate; and (iii) a Bernstein-polynomial graph spectral filter with a claimed BIBO-stability guarantee and symmetric channel mixing for spectral adaptation. The authors report supervised and unsupervised experiments on ShapeNet, Stanford, and Paris-rue-Madame data, with CD, EMD, HD, and RMSD metrics, and they provide ablation studies comparing the proposed graph construction and spectral filtering against simplified variants. The central claim, stated in the abstract and Section I, is that GD-GCN outperforms state-of-the-art denoising methods through these three mechanisms.

Significance. If the technical gaps were repaired, GD-GCN would be a useful contribution: the neural-ODE perspective on dynamic graph construction is timely, the experimental campaign is broad (synthetic and LiDAR data, supervised and unsupervised settings), and the reported CD/EMD gains in Tables II-V are substantial. The paper also includes an ablation study that isolates the proposed components, which is a strength. However, the Riemannian-metric derivation and the BIBO-stability proof contain load-bearing errors, and the empirical claims are presented without error bars or code. The significance is therefore conditional on correcting the theoretical claims and strengthening the experimental evidence.

major comments (4)
  1. [Section III-D2, Eq. (8)] The learned object J_theta is not a Jacobian in any stated sense. With X_tilde in R^{N x K(d+d_l)}, the product (X_tilde W_Q)(X_tilde W_K)^T is N x N before the concatenation and linear projection to R^{N x d}; there is no per-point d x d_l matrix. Moreover, Eq. (7) is dimensionally inconsistent with the stated map z : R^{d_l} -> R^d: (nabla z(p))^T (nabla z(p)) is d_l x d_l, not d x d. Consequently, the 'Riemannian distance' in Eq. (6) is at best a learned quadratic form, and the ablation in Section IV-D2 cannot establish the geometric claim. The authors should either provide a correct derivation of a per-point metric from a bona fide Jacobian, or explicitly downgrade the contribution to a learned Mahalanobis-style distance.
  2. [Proposition 1 and Appendix A] The coefficient formula in Proposition 1 contradicts the normalization used in the proof. Taking g equiv 1 and K = 2 gives theta_k = 2^{-2} * 3! / 1! = 3/2 for k = 0, 1, 2; then B_2(lambda) = (3/2)[(1-lambda)^2 + 2 lambda (1-lambda) + lambda^2] = 3/2, so B_2(0.5) = 1.5 > 1, violating the claimed bound 0 < B_K(lambda) <= 1. The proof's normalization factor (K - floor(K/2))! / (K+1)! * 2^K is inverted relative to the stated theta_k; the correct coefficient would multiply by 2^K (K - floor(K/2))! / (K+1)!, not divide by it. As stated, the BIBO-stability guarantee is false and must be corrected.
  3. [Section III-E2, Eq. (11)] The eigenvalue identity for channel mixing is dimensionally inconsistent. With mu, phi in R^d (eigenvalues of the d x d mixing matrices) and phi in R^N (eigenvalues of B_K), the eigenvalues of a Kronecker-sum operator acting on the vectorized N x d state lie in R^{dN} and are of the form mu_i phi_j - phi_k (or, under additional commuting assumptions, mu_i phi_j - phi_i), i.e., mu ⊗ phi - 1 ⊗ phi in the commuting case. The expression mu ⊗ phi - phi ⊗ 1 combines vectors of lengths d^2 and dN and cannot be the spectrum of the stated operator. Please correct the identity and the subsequent 'outer product' description, and state any simultaneous-diagonalizability assumptions needed for the Kronecker-sum eigenvalue factorization.
  4. [Section IV, Tables II-V] The central empirical claim rests on Tables II-V, but no error bars, confidence intervals, or significance tests are provided, and the description of baseline training ('train the remaining models to achieve best performance', Section IV-A3) makes it difficult to assess whether the reported gains are robust or favorable to the proposed method. Reporting variance over at least a few random seeds or test subsets is necessary to support the 'outperforms state-of-the-art' conclusion, especially where the reported differences are small (e.g., the dense 30K cases at several noise levels in Table II).
minor comments (6)
  1. [Abstract] The abstract contains an incomplete sentence: 'fitting the point cloud with noise to the underlying surface by and the learning process for MST-GConv acts like a changing system' should be rephrased to remove the dangling 'by'.
  2. [Section II-B] The word 'categoried' should be 'categorized'.
  3. [Section III-C2] The sentence 'Subsequently, we the framework also updates A_t^G in real-time' is grammatically incomplete; it should be 'Subsequently, the framework also updates A_t^G in real-time'.
  4. [Section III-F1, Eqs. (13)-(14)] The notation in the loss definitions is inconsistent: Eq. (13) mixes u_j and the denoised point variable in the same expression, and Eq. (14) writes a sum over points in the set of denoised points using an unclear subscript. Please align the indices carefully.
  5. [References] References [28] and [68] are the same paper (Wang et al., 'Dynamic graph CNN for learning on point clouds') listed twice with different numbers; please merge the duplicate citation.
  6. [Proposition 1] The proposition states 'for any k in N' but k ranges over 0, ..., K; the quantifier should be 'for k = 0, 1, ..., K'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's claims are supported by held-out empirical evaluation and by design constraints, not by a fitted-parameter or self-citation chain.

full rationale

The paper's central claim is empirical: GD-GCN is compared against ten baselines on held-out ShapeNet, Stanford, and Paris-rue-Madame data using CD, EMD, and HD, with results reported in Tables II–V. The architecture is specified by equations (3)–(10) with learnable parameters trained by the losses in (12)–(17). No quantity fitted on a subset is later reported as a prediction of a closely related quantity; using CD and EMD as both training losses and evaluation metrics is standard supervised practice and does not force the outcome. The Bernstein stability result is a coefficient constraint, not a derived consequence of the reported performance: Proposition 1 and Appendix A show that normalizing coefficients bounds B_K(λ) by 1, and that bound is imposed before training, not extracted from the data. The Riemannian metric construction in equations (7)–(8) does raise a serious validity concern, because the attention output J_θ has shape N×d (or N×N before projection) while the Jacobian of the map z: R^{d_l} → R^d would be d×d_l, so identifying J_θ with ∇z(p) is unsupported; however, this is a correctness or interpretability issue, not circularity, because the network is not defined in terms of its own outputs and the ablation 'GD-GCN w/o Geo-Graph' compares two genuinely different graph constructions on held-out data. No load-bearing premise rests on a self-citation: references such as BernNet [52] and GRAND [57] are external prior work, and no uniqueness theorem or fitted ansatz is imported from the authors' own earlier papers. Therefore no circular step is identifiable in the derivation chain.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The method rests on standard manifold and approximation hypotheses plus one ad hoc identification: the attention output is treated as a Jacobian. The central empirical claim also depends on typical deep learning hyperparameters, none of which are derived.

free parameters (6)
  • scale factor alpha = not reported
    Controls the relative weight of spatial coordinates and features in the embedding X = (P, alpha Z) used to compute the Riemannian metric; hand-chosen but not specified.
  • ODE time step Delta t = 0.1
    Fixed-step RK4 solver step for Eq. (5); chosen by hand.
  • terminal time T = 1
    Integration horizon in Eq. (5); chosen by hand.
  • kNN neighborhood k = 16
    Number of neighbors in graph construction; chosen by hand.
  • repulsion loss weight lambda = 0.01
    Weight of the repulsion loss in unsupervised training, Eq. (15); chosen by hand.
  • Bernstein polynomial order K = not specified
    Order of the Bernstein approximation in Eq. (9); affects filter flexibility and stability but is not reported.
assumptions (5)
  • standard math Bernstein polynomial approximation converges to continuous functions as K increases.
    Used in Eq. (9) and Section III-E1, citing [60].
  • domain assumption 3-D point clouds reside on a low-dimensional Riemannian manifold.
    Invoked in Section III-D1 via the Riemannian manifold hypothesis [54].
  • domain assumption Noise is additive Gaussian with standard deviation in [0.01, 0.03].
    Training and synthetic test setup in Section IV-A2; the real LiDAR test uses Blensor noise.
  • ad hoc to paper The multi-head attention output J_theta is a valid approximation of the Jacobian gradient z(p).
    Eq. (8) substitutes J_theta for gradient z(p) in Eq. (7) with no derivation or error bound; dimensions are inconsistent.
  • domain assumption Graph filter eigenvalues lie in (0, 1] as required by Proposition 1.
    Proposition 1 states lambda in (0, 1]; the paper does not specify the normalization of the adjacency matrix that would guarantee this range.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Point Cloud Denoising With Fine-Granularity Dynamic Graph Convolutional Networks." pith.science (2026). https://pith.science/paper/QWMTCP3L

@misc{pith2026241114158,
  author       = {Pith},
  title        = {Pith review of: Point Cloud Denoising With Fine-Granularity Dynamic Graph Convolutional Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QWMTCP3L}},
  note         = {Machine review of arXiv:2411.14158}
}
read the original abstract

Due to limitations in acquisition equipment, noise perturbations often corrupt 3-D point clouds, hindering down-stream tasks such as surface reconstruction, rendering, and further processing. Existing 3-D point cloud denoising methods typically fail to reliably fit the underlying continuous surface, resulting in a degradation of reconstruction performance. This paper introduces fine-granularity dynamic graph convolutional networks called GD-GCN, a novel approach to denoising in 3-D point clouds. The GD-GCN employs micro-step temporal graph convolution (MST-GConv) to perform feature learning in a gradual manner. Compared with the conventional GCN, which commonly uses discrete integer-step graph convolution, this modification introduces a more adaptable and nuanced approach to feature learning within graph convolution networks. It more accurately depicts the process of fitting the point cloud with noise to the underlying surface by and the learning process for MST-GConv acts like a changing system and is managed through a type of neural network known as neural Partial Differential Equations (PDEs). This means it can adapt and improve over time. GD-GCN approximates the Riemannian metric, calculating distances between points along a low-dimensional manifold. This capability allows it to understand the local geometric structure and effectively capture diverse relationships between points from different geometric regions through geometric graph construction based on Riemannian distances. Additionally, GD-GCN incorporates robust graph spectral filters based on the Bernstein polynomial approximation, which modulate eigenvalues for complex and arbitrary spectral responses, providing theoretical guarantees for BIBO stability. Symmetric channel mixing matrices further enhance filter flexibility by enabling channel-level scaling and shifting in the spectral domain.

Figures

Figures reproduced from arXiv: 2411.14158 by the authors.

Figure 1
Figure 1. (a) Evolving trajectory of the noise points relative to the underlying surface by coarse-grained dynamics GCN. (b) Evolving trajectory of the noise [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The progression of point cloud denoising. (a) is the noisy input, (b) [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. (a) Overall architecture of proposed fine-grained dynamics graph convolution networks (GD-GCN). (b) The details of Riemannian metric [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Denoising results for airplane and car from the ShapeNet repository. From top left to bottom right: Groud Truth, GPDNet [27], TD [22], DMR [23], ScoreNet [24], Deeprs [69], and PathNet [70] and the proposed method. CD values in 10−6 are reported. and is elegantly modif…
Figure 5
Figure 5. Figure 5: Denoising results for the bunny and dragon from the Stanford repository using both supervised (top) and unsupervised (bottom) modes. From top left to bottom right: TD [22], DMR [23], ScoreNet [24], ours and ground truth. CD values in 10−6 are reported [PITH_FULL_IMAGE…
Figure 6
Figure 6. Figure 6: Denoising real data results for real-world data with LiDAR noise. From left to right: TD [22], DMR [23], ScoreNet [24], ours and ground truth. The [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Ablation experiment results. The performance curves of DTL-GCN, GD-GCN w/o Geo-Graph, GD-GCN w/o Spectral-Filtering, and Proposed in the [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Outcomes of the denoising process achieved through geometric graph [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: Dynamic denoising analysis of MST-GConv. Different termination times lead to Different denoising results, where orange represents denoised points [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Deep Learning For Point Cloud Denoising: A Survey

    cs.CV 2025-08 unverdicted novelty 4.0 of 10

    A survey of deep learning point cloud denoising, proposing a taxonomy of outlier removal and surface restoration.

Reference graph

Works this paper leans on

76 extracted references · 72 canonical work pages · cited by 1 Pith paper

  1. [1]

    Temporal-channel transformer for 3D Lidar-based video object detection for autonomous driving,

    Z. Yuan, X. Song, L. Bai, Z. Wang, and W. Ouyang, “Temporal-channel transformer for 3D Lidar-based video object detection for autonomous driving,” IEEE Trans. Circuits Syst. Video Technol. , vol. 32, no. 4, pp. 2068–2078, 2022

    work page 2022

  2. [2]

    Transformer3D-Det: Improving 3D object detection by vote refinement,

    L. Zhao, J. Guo, D. Xu, and L. Sheng, “Transformer3D-Det: Improving 3D object detection by vote refinement,” IEEE Trans. Circuits Syst. Video Technol., vol. 31, no. 12, pp. 4735–4746, 2021

    work page 2021

  3. [3]

    UrbanLF: A comprehensive light field dataset for semantic segmentation of urban scenes,

    H. Sheng, R. Cong, D. Yang, R. Chen, S. Wang, and Z. Cui, “UrbanLF: A comprehensive light field dataset for semantic segmentation of urban scenes,” IEEE Trans. Circuits Syst. Video Technol. , vol. 32, no. 11, pp. 7880–7893, 2022. 13 Fig. 9. Dynamic denoising analysis of MST-GConv. Different termination times lead to Different denoising results, where oran...

    work page 2022

  4. [4]

    Progres- sive point cloud upsampling via differentiable rendering,

    P. Zhang, X. Wang, L. Ma, S. Wang, S. Kwong, and J. Jiang, “Progres- sive point cloud upsampling via differentiable rendering,” IEEE Trans. Circuits Syst. Video Technol., vol. 31, no. 12, pp. 4673–4685, 2021

    work page 2021

  5. [5]

    Point cloud upsampling via per- turbation learning,

    D. Ding, C. Qiu, F. Liu, and Z. Pan, “Point cloud upsampling via per- turbation learning,” IEEE Trans. Circuits Syst. Video Technol. , vol. 31, no. 12, pp. 4661–4672, 2021

    work page 2021

  6. [6]

    The bilateral filter for point clouds,

    J. Digne and C. De Franchis, “The bilateral filter for point clouds,” Image Process. On Line , vol. 7, pp. 278–287, 2017

    work page 2017

  7. [7]

    Point cloud denoising with principal component analysis and a novel bilateral filter,

    F. Zhang, C. Zhang, H. Yang, and L. Zhao, “Point cloud denoising with principal component analysis and a novel bilateral filter,” Traitement du Signal, vol. 36, no. 5, pp. 393–398, 2019

    work page 2019

  8. [8]

    Computing and rendering point set surfaces,

    M. Alexa, J. Behr, D. Cohen-Or, S. Fleishman, D. Levin, and C. T. Silva, “Computing and rendering point set surfaces,” IEEE Trans. Vis. Comput. Graphics, vol. 9, no. 1, pp. 3–15, 2003

    work page 2003

Show all 76 references
  1. [9]

    Feature preserving point set surfaces based on non-linear kernel regression,

    A. C. ¨Oztireli, G. Guennebaud, and M. Gross, “Feature preserving point set surfaces based on non-linear kernel regression,” Comput. Graph. Forum, vol. 28, no. 2, pp. 493–501, 2009

    2009

  2. [10]

    Algebraic point set surfaces,

    G. Guennebaud and M. Gross, “Algebraic point set surfaces,” ACM Trans. Graph., vol. 26, no. 3, article no. 23, pp. 1–10, 2007

    2007

  3. [11]

    Parameterization- free projection for geometry reconstruction,

    Y . Lipman, D. Cohen-Or, D. Levin, and H. Tal-Ezer, “Parameterization- free projection for geometry reconstruction,” ACM Trans. Graph. , vol. 26, no. 3, article no. 22, pp. 1–6, 2007

    2007

  4. [12]

    Edge-aware point set resampling,

    H. Huang, S. Wu, M. Gong, D. Cohen-Or, U. Ascher, and H. Zhang, “Edge-aware point set resampling,” ACM Trans. Graph., vol. 32, no. 1, article no. 9, pp. 1–12, 2013

    2013

  5. [13]

    Estimating differential quantities using poly- nomial fitting of osculating jets,

    F. Cazals and M. Pouget, “Estimating differential quantities using poly- nomial fitting of osculating jets,” Comput. Aided Geom. Des. , vol. 22, no. 2, pp. 121–146, 2005

    2005

  6. [14]

    ℓ1-sparse reconstruc- tion of sharp point set surfaces,

    H. Avron, A. Sharf, C. Greif, and D. Cohen-Or, “ ℓ1-sparse reconstruc- tion of sharp point set surfaces,” ACM Trans. Graph. , vol. 29, no. 5, article no. 135, pp. 1–12, 2010

    2010

  7. [15]

    Denoising point sets via L0 minimization,

    Y . Sun, S. Schaefer, and W. Wang, “Denoising point sets via L0 minimization,” Comput. Aided Geom. Des. , vol. 35, pp. 2–15, 2015

    2015

  8. [16]

    Point cloud denoising via moving RPCA,

    E. Mattei and A. Castrodad, “Point cloud denoising via moving RPCA,” Comput. Graph. Forum, vol. 36, no. 8, pp. 123–137, 2017

    2017

  9. [17]

    3D point cloud denoising using graph laplacian regularization of a low dimensional manifold model,

    J. Zeng, G. Cheung, M. Ng, J. Pang, and C. Yang, “3D point cloud denoising using graph laplacian regularization of a low dimensional manifold model,” IEEE Trans. Image Process., vol. 29, pp. 3474–3489, 2019

    2019

  10. [18]

    3D point cloud denoising via bipartite graph approximation and reweighted graph Laplacian,

    C. Dinesh, G. Cheung, and I. V . Bajic, “3D point cloud denoising via bipartite graph approximation and reweighted graph Laplacian,” arXiv preprint arXiv:1812.07711, 2018

    2018 arXiv

  11. [19]

    Graph-based denoising for time-varying point clouds,

    Y . Schoenenberger, J. Paratte, and P. Vandergheynst, “Graph-based denoising for time-varying point clouds,” in 2015 3DTV-Conf.: True Vis. - Capture, Transmiss. Display 3D Video (3DTV-CON) , 2015

    2015

  12. [20]

    PointCleanNet: Learning to denoise and remove outliers from dense point clouds,

    M.-J. Rakotosaona, V . La Barbera, P. Guerrero, N. J. Mitra, and M. Ovs- janikov, “PointCleanNet: Learning to denoise and remove outliers from dense point clouds,” Comput. Graph. Forum, vol. 39, no. 1, pp. 185–203, 2020

    2020

  13. [21]

    3D point cloud denoising via deep neural network based local surface estimation,

    C. Duan, S. Chen, and J. Kovacevic, “3D point cloud denoising via deep neural network based local surface estimation,” in 2019 IEEE Int. Conf. Acoust., Speech Signal Process. (ICASSP) , 2019, pp. 8553–8557

    2019

  14. [22]

    Total denoising: Unsuper- vised learning of 3D point cloud cleaning,

    P. Hermosilla, T. Ritschel, and T. Ropinski, “Total denoising: Unsuper- vised learning of 3D point cloud cleaning,” in 2019 IEEE/CVF Int. Conf. Comput. Vis. (ICCV) , 2019, pp. 52–60

    2019

  15. [23]

    Differentiable manifold reconstruction for point cloud denoising,

    S. Luo and W. Hu, “Differentiable manifold reconstruction for point cloud denoising,” in Proc. 28th ACM Int. Conf. Multimedia , 2020, pp. 1330–1338

    2020

  16. [24]

    Score-based point cloud denoising,

    ——, “Score-based point cloud denoising,” in 2021 IEEE/CVF Int. Conf. Comput. Vis. (ICCV) , 2021, pp. 4583–4592

    2021

  17. [25]

    3D point cloud color denoising using convex graph-signal smoothness priors,

    C. Dinesh, G. Cheung, and I. Baji ´c, “3D point cloud color denoising using convex graph-signal smoothness priors,” in 2019 IEEE 21st Int. Workshop Multimedia Signal Process. (MMSP) , 2019, pp. 1–6

    2019

  18. [26]

    PointNet: Deep learning on point sets for 3D classification and segmentation,

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas, “PointNet: Deep learning on point sets for 3D classification and segmentation,” in 2017 IEEE Conf. Comput. Vis. Pattern Recognit. , 2017, pp. 652–660

    2017

  19. [27]

    Learning graph- convolutional representations for point cloud denoising,

    F. Pistilli, G. Fracastoro, D. Valsesia, and E. Magli, “Learning graph- convolutional representations for point cloud denoising,” in 16th Eur. Conf. Comput. Vis. (ECCV) , 2020, pp. 103–118

    2020

  20. [29]

    Geometric deep learning: Going beyond Euclidean data,

    M. M. Bronstein, J. Bruna, Y . LeCun, A. Szlam, and P. Vandergheynst, “Geometric deep learning: Going beyond Euclidean data,” IEEE Signal Process. Mag., vol. 34, no. 4, pp. 18–42, 2017

    2017

  21. [30]

    The emerging field of signal processing on graphs: Ex- tending high-dimensional data analysis to networks and other irregular domains,

    D. I. Shuman, S. K. Narang, P. Frossard, A. Ortega, and P. Van- dergheynst, “The emerging field of signal processing on graphs: Ex- tending high-dimensional data analysis to networks and other irregular domains,” IEEE Signal Process. Mag. , vol. 30, no. 3, pp. 83–98, 2013

    2013

  22. [31]

    The curse of dimensionality,

    M. K ¨oppen, “The curse of dimensionality,” in 5th Online World Conf. Soft Comput. Industr. Appl. (WSC5) , vol. 1, 2000, pp. 4–8

    2000

  23. [32]

    Bilateral filtering for gray and color images,

    C. Tomasi and R. Manduchi, “Bilateral filtering for gray and color images,” in 6th Int. Conf. Comput. Vis. , 1998, pp. 839–846

    1998

  24. [33]

    PointNet++: Deep hierarchical feature learning on point sets in a metric space,

    C. R. Qi, L. Yi, H. Su, and L. J. Guibas, “PointNet++: Deep hierarchical feature learning on point sets in a metric space,” in Adv. Neural Inf. Process. Syst. 30 , 2017, pp. 5099–5108

    2017

  25. [34]

    PCPNet learning local shape properties from raw point clouds,

    P. Guerrero, Y . Kleiman, M. Ovsjanikov, and N. J. Mitra, “PCPNet learning local shape properties from raw point clouds,” Comput. Graph. Forum, vol. 37, no. 2, pp. 75–85, 2018

    2018

  26. [35]

    PointFilterNet: A filtering network for point cloud denoising,

    X. Wang, X. Fan, and D. Zhao, “PointFilterNet: A filtering network for point cloud denoising,” IEEE Trans. Circuits Syst. Video Technol. , vol. 33, no. 3, pp. 1276–1290, 2023

    2023

  27. [36]

    FCNet: Learning noise-free features for point cloud denoising,

    X. Wang, W. Cui, R. Xiong, X. Fan, and D. Zhao, “FCNet: Learning noise-free features for point cloud denoising,” IEEE Trans. Circuits Syst. Video Technol., vol. 33, no. 11, pp. 6288–6301, 2023

    2023

  28. [37]

    PointProNets: Consolidation of point clouds with convolutional neural networks,

    R. Roveri, A. C. ¨Oztireli, I. Pandele, and M. Gross, “PointProNets: Consolidation of point clouds with convolutional neural networks,” Comput. Graph. Forum, vol. 37, no. 2, pp. 87–99, 2018

    2018

  29. [38]

    Learning robust graph-convolutional representations for point cloud denoising,

    F. Pistilli, G. Fracastoro, D. Valsesia, and E. Magli, “Learning robust graph-convolutional representations for point cloud denoising,” IEEE J. Sel. Topics Signal Process. , vol. 15, no. 2, pp. 402–414, 2020

    2020

  30. [39]

    Semi-supervised classification with graph convolutional networks,

    T. N. Kipf and M. Welling, “Semi-supervised classification with graph convolutional networks,” in 5th Int. Conf. Learn. Rep. , 2017

    2017

  31. [40]

    Beltrami flow and neural diffusion on graphs,

    B. Chamberlain, J. Rowbottom, D. Eynard, F. Di Giovanni, X. Dong, and M. Bronstein, “Beltrami flow and neural diffusion on graphs,” in Adv. Neural Inf. Process. Syst. 34 , pp. 1594–1609, 2021. 14

    2021

  32. [41]

    Neural ordinary differential equations,

    R. T. Q. Chen, Y . Rubanova, J. Bettencourt, and D. Duvenaud, “Neural ordinary differential equations,” in Adv. Neural Inf. Process. Syst. 31 , pp. 6571–6583, 2018

    2018

  33. [42]

    Augmented neural ODEs,

    E. Dupont, A. Doucet, and Y . W. Teh, “Augmented neural ODEs,” in Adv. Neural Inf. Process. Syst. 32 , 2019, pp. 3140–3150

    2019

  34. [43]

    Graph neural ordinary differential equations,

    M. Poli, S. Massaroli, J. Park, A. Yamashita, H. Asama, and J. Park, “Graph neural ordinary differential equations,” arXiv preprint arXiv:1911.07532, 2019

    1911 arXiv

  35. [44]

    A proposal on machine learning via dynamical systems,

    E. Weinan, “A proposal on machine learning via dynamical systems,” Commun. Math. Stat. , vol. 1, no. 5, pp. 1–11, 2017

    2017

  36. [45]

    Beyond finite layer neural net- works: Bridging deep architectures and numerical differential equations,

    Y . Lu, A. Zhong, Q. Li, and B. Dong, “Beyond finite layer neural net- works: Bridging deep architectures and numerical differential equations,” in Proc. 35th Int. Conf. Mach. Learn. , 2018, pp. 3276–3285

    2018

  37. [46]

    Predict then propagate: Graph neural networks meet personalized PageRank,

    J. Gasteiger, A. Bojchevski, and S. G ¨unnemann, “Predict then propagate: Graph neural networks meet personalized PageRank,” in 7th Int. Conf. Learn. Rep., 2019

    2019

  38. [47]

    Interpreting and unifying graph neural networks with an optimization framework,

    M. Zhu, X. Wang, C. Shi, H. Ji, and P. Cui, “Interpreting and unifying graph neural networks with an optimization framework,” in Proc. Web Conf. 2021 (WWW‘21) , 2021, pp. 1215–1226

    2021

  39. [48]

    Representation learning on graphs with jumping knowledge networks,

    K. Xu, C. Li, Y . Tian, T. Sonobe, K.-i. Kawarabayashi, and S. Jegelka, “Representation learning on graphs with jumping knowledge networks,” in Proc. 35th Int. Conf. Mach. Learn. , 2018, pp. 5453–5462

    2018

  40. [49]

    Towards deeper graph neural networks,

    M. Liu, H. Gao, and S. Ji, “Towards deeper graph neural networks,” in Proc. 26th ACM SIGKDD Int. Conf. Knowl. Disc. Data Min. , 2020, pp. 338–348

    2020

  41. [50]

    Convolutional neural networks on graphs with fast localized spectral filtering,

    M. Defferrard, X. Bresson, and P. Vandergheynst, “Convolutional neural networks on graphs with fast localized spectral filtering,” in Adv. Neural Inf. Process. Syst. 29 , 2016, pp. 3844–3852

    2016

  42. [51]

    Graph neural networks with convolutional ARMA filters,

    F. M. Bianchi, D. Grattarola, L. Livi, and C. Alippi, “Graph neural networks with convolutional ARMA filters,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 44, no. 7, pp. 3496–3507, 2021

    2021

  43. [52]

    BernNet: Learning arbitrary graph spectral filters via Bernstein approximation,

    M. He, Z. Wei, Z. Huang, and H. Xu, “BernNet: Learning arbitrary graph spectral filters via Bernstein approximation,” in Adv. Neural Inf. Process. Syst. 34 , 2021, pp. 14 239–14 251

    2021

  44. [53]

    Images as embedded maps and minimal surfaces: Movies, color, texture, and volumetric medical images,

    R. Kimmel, R. Malladi, and N. Sochen, “Images as embedded maps and minimal surfaces: Movies, color, texture, and volumetric medical images,” Int. J. Comput. Vis. , vol. 39, no. 2, pp. 111–129, 2000

    2000

  45. [54]

    J. M. Lee, Introduction to Riemannian Manifolds , 2nd ed. Springer Cham, 2018

    2018

  46. [55]

    The existence of complete Riemannian metric,

    K. Nomizu and H. Ozeki, “The existence of complete Riemannian metric,” Proc. Am. Math. Soc. , vol. 12, no. 6, pp. 889–891, 1961

    1961

  47. [56]

    Attention is all you need,

    A. Vaswani et al., “Attention is all you need,” inAdv. Neural Inf. Process. Syst. 30, 2017, pp. 5998–6008

    2017

  48. [57]

    Grand: Graph neural diffusion,

    B. Chamberlain, J. Rowbottom, M. I. Gorinova, M. Bronstein, S. Webb, and E. Rossi, “Grand: Graph neural diffusion,” in Proc. 38th Int. Conf. Mach. Learn., 2021, pp. 1407–1418

    2021

  49. [58]

    Continuous graph neural net- works,

    L.-P. Xhonneux, M. Qu, and J. Tang, “Continuous graph neural net- works,” in Proc. 37th Int. Conf. Mach. Learn. , 2020, pp. 10 432–10 441

    2020

  50. [59]

    PairNorm: Tackling oversmoothing in gnns,

    L. Zhao and L. Akoglu, “PairNorm: Tackling oversmoothing in gnns,” in 8th Int. Conf. Learn. Rep. , 2020

    2020

  51. [60]

    The Bernstein polynomial basis: A centennial retrospec- tive,

    R. T. Farouki, “The Bernstein polynomial basis: A centennial retrospec- tive,” Comput. Aided Geom. Des. , vol. 29, no. 6, pp. 379–419, 2012

    2012

  52. [61]

    Guided image filtering,

    K. He, J. Sun, and X. Tang, “Guided image filtering,” in 11th Eur. Conf. Comput. Vis. (ECCV) , 2010, pp. 1–14

    2010

  53. [62]

    A point set generation network for 3D object reconstruction from a single image,

    H. Fan, H. Su, and L. J. Guibas, “A point set generation network for 3D object reconstruction from a single image,” in 2017 IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR) , 2017, pp. 605–613

    2017

  54. [63]

    ShapeNet: An information-rich 3D model reposi- tory,

    A. X. Chang et al., “ShapeNet: An information-rich 3D model reposi- tory,” arXiv preprint arXiv:1512.03012 , 2015

    2015 arXiv

  55. [64]

    VisionAir, http://www.infra-visionair.eu/ Accessed November 14, 2017

    2017

  56. [65]

    Linear light source reflectometry,

    A. Gardner, C. Tchou, T. Hawkins, and P. Debevec, “Linear light source reflectometry,” ACM Trans. Graph., vol. 22, no. 3, pp. 749–758, 2003

    2003

  57. [66]

    Paris-rue- Madame database: a 3D mobile laser scanner dataset for benchmarking urban detection, segmentation and classification methods,

    A. Serna, B. Marcotegui, F. Goulette, and J.-E. Deschaud, “Paris-rue- Madame database: a 3D mobile laser scanner dataset for benchmarking urban detection, segmentation and classification methods,” in Proc. 3rd Int. Conf. Pattern Recognit. Appl. Methods , 2014, pp. 819–824

    2014

  58. [67]

    BlenSor: Blender sensor simulation toolbox,

    M. Gschwandtner, R. Kwitt, A. Uhl, and W. Pree, “BlenSor: Blender sensor simulation toolbox,” in Int. Symp. Vis. Comput. , 2011, pp. 199– 208

    2011

  59. [68]

    Dynamic graph CNN for learning on point clouds,

    Y . Wang, Y . Sun, Z. Liu, S. E. Sarma, M. M. Bronstein, and J. M. Solomon, “Dynamic graph CNN for learning on point clouds,” ACM Trans. Graph., vol. 38, no. 5, article no. 146, pp. 1–12, 2019

    2019

  60. [69]

    Deep point set resampling via gradient fields,

    H. Chen, S. Luo, and W. Hu, “Deep point set resampling via gradient fields,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 45, no. 3, pp. 2913– 2930, 2022

    2022

  61. [70]

    PathNet: Path-selective point cloud denoising,

    Z. Wei, H. Chen, L. Nan, J. Wang, J. Qin, and M. Wei, “PathNet: Path-selective point cloud denoising,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 46, no. 6, pp. 4426–4442, 2024

    2024

  62. [71]

    A comprehensive survey on graph neural networks,

    Z. Wu, S. Pan, F. Chen, G. Long, C. Zhang, and S. P. Yu, “A comprehensive survey on graph neural networks,” IEEE Trans. Neural Netw. Learn. Syst. , vol. 32, no. 1, pp. 4–24, 2020

    2020

  63. [72]

    Deeper insights into graph convolutional networks for semi-supervised learning,

    Q. Li, Z. Han, and X.-M. Wu, “Deeper insights into graph convolutional networks for semi-supervised learning,” in Proc. 32nd AAAI Conf. Artif. Intell., 2018, pp. 3538–3545

    2018

  64. [73]

    LanczosNet: Multi- scale deep graph convolutional networks,

    R. Liao, Z. Zhao, R. Urtasun, and R. S. Zemel, “LanczosNet: Multi- scale deep graph convolutional networks,” in 7th Int. Conf. Learn. Rep. , 2019

    2019

  65. [74]

    Regularization on discrete spaces,

    D. Zhou and B. Sch ¨olkopf, “Regularization on discrete spaces,” in Proc. Joint Pattern Recognit. Symp. , 2005, pp. 361–368

    2005

  66. [75]

    Understanding convolution on graphs via energies,

    F. Di Giovanni, J. Rowbottom, B. P. Chamberlain, T. Markovich, and M. M. Bronstein, “Understanding convolution on graphs via energies,” Trans. Mach. Learn. Res. , 2023

    2023

  67. [76]

    A note on over-smoothing for graph neural networks,

    C. Cai and Y . Wang, “A note on over-smoothing for graph neural networks,” arXiv preprint arXiv:2006.13318 , 2020

    2006 arXiv

  68. [77]

    BIBO Stability and Decomposition Analysis of Signals and System with Convolution Techniques,

    C. B. Sumathi and R. Jothilakshmi, “BIBO Stability and Decomposition Analysis of Signals and System with Convolution Techniques,” Ratio Mathematica, vol. 46, no. 0, 2023

    2023

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.