Pith. sign in

REVIEW 3 major objections 5 minor 59 references

Open Your Eyes: Vision Enhances Message Passing Neural Networks in Link Prediction

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Vision features extracted from rendered subgraph images improve MPNN link prediction across all seven benchmarks and set new state-of-the-art results.

desk verdict The paper's gains likely come from the queried edge remaining in the visual layout; the leak must be fixed before the claim is credible. read the letter →

arxiv 2505.08266 v3 pith:DCPROB6T submitted 2025-05-13 cs.CV cs.AIcs.LG

Yanbin Wei , Xuehao Wang , Zhan Zhuang , Yang Chen , Shuhao Chen , Yulong Zhang , Yu Zhang , James Kwok This is my paper · ORCID
classification cs.CVcs.AIcs.LG
keywords linkpredictionmessagepassingneuralnetworksvisualstructuralfeaturessubgraphvisualizationvisionencodergraphmultimodallearning
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's central claim is that a message-passing link predictor can see better when it literally looks: render the k-hop subgraph around each candidate link as a consistent-style image, read that image with a pretrained vision encoder, and fuse the resulting visual structural features into the MPNN. On this basis the authors propose the Graph Vision Network (GVN) and an efficient variant (E-GVN), and report that the visual channel improves link prediction across all seven datasets tested, including four large-scale OGB benchmarks, while remaining a plug-in compatible with existing state-of-the-art models. If the claim holds, vision becomes an orthogonal source of structural signal, capable of distinguishing links that message passing and hand-designed structural features cannot, and a broad new direction for graph learning opens up.

What carries the argument

The load-bearing object is the Visual Structural Feature (VSF): the output vector of a vision encoder (default ResNet50) applied to a fixed-style image of the k-hop subgraph enclosing the candidate link (GVN) or centered on a node (E-GVN), with the queried endpoints colored, labels removed, and the link itself masked. The rendering pipeline, subgraph extraction, Graphviz sfdp layout, and consistent style, converts topology into pixels in a way the encoder can read; the visual perception scope $k \le 3$ is deliberately decoupled from message-passing depth, so the model sees a crisp local picture while the MPNN reaches wider. The framework's other moving parts are the three integration strategies (attention-based, concatenated, and weighted) that inject VSFs without touching the message-passing loop, and, in E-GVN, the node-centered visualization that cuts rendering cost from $\mathcal{O}(l)$ per-link to $\mathcal{O}(n)$ per-node plus a frozen-encoder/adaptor design that keeps memory flat.

What would settle it

Run GVN with all rendering settings fixed except the layout: replace the force-directed sfdp placement with random node positions on the same canvas. If link-prediction accuracy stays at the same level, the encoder is reading pixel artifacts rather than structure and the gains would not generalize across renderers; if accuracy collapses, layout fidelity is load-bearing. A second check: replace the pretrained ResNet50 with a randomly initialized encoder of the same architecture; if the gains survive, natural-image pretraining is not the source of the VSF signal.

Watch

Extended reading notes

Core claim

GVN and E-GVN establish, empirically, that visual structural features (VSFs), vectors produced by a ResNet50 encoder applied to a force-directed rendering of a k-hop subgraph, carry genuine link-relevant information that MPNN representations and classical structural features do not fully cover. The paper demonstrates this three ways: link-centered images of subgraphs with isomorphic endpoints are visibly different, so VSFs discriminate links that the 1-WL-limited MPNN equates; VSFs let plain GCN and SAGE count triangles and 3-stars with near-zero normalized error in synthetic substructure-counting tests; and VSFs reproduce the values of six standard structural features (CN, RA, AA, SPD, DRNL, DE), with the reproduced mix shifting by dataset density after finetuning. Integrated into MPNNs by cross-attention, concatenation, or weighted prediction, VSFs improve GCN by 21 to 38 percent relative on the citation networks and push the state-of-the-art model NCNC higher on every one of the seven datasets, which the paper reads as evidence that vision awareness is orthogonal to common-neighbor and path-based features. E-GVN moves rendering from per-link to per-node, freezes the encoder, and appends a trainable adapter, which brings the same benefits to large-scale graphs at near-base-model cost.

Load-bearing premise

A force-directed drawing of a small subgraph, read by an image encoder pretrained on natural photographs, preserves enough of the topology that matters for links (shared neighbors, distances, motifs) that the encoded features genuinely help predict links rather than just re-encoding drawing artifacts.

Editorial extensions

If this is right

  • Vision enhancement lifts plain MPNNs: GVN and E-GVN over GCN give relative HR@100 gains of about 21%, 25%, and 38% on Cora, Citeseer, and PubMed respectively.
  • The gains stack on top of the strongest common-neighbor-based model: GVN and E-GVN with NCNC beat NCNC on all seven datasets, both on the headline metric and on most of the broader hit@k and MRR metrics.
  • VSFs supply MPNNs with substructure-counting ability: adding VSFs drops normalized counting error for triangles and 3-stars from order-1 values to below $10^{-5}$ in the synthetic tests.
  • Because the visual scope is decoupled from message-passing depth, the model gets both a refined local structural readout and a wider message-passing reach, so the design does not sacrifice depth for vision.
  • E-GVN's node-centered, frozen-encoder design makes the approach usable on large-scale graphs: relative gains of 38.86% (HR@50) on ogbl-collab, 72.20% (HR@100) on ogbl-ppa, 1.60% (MRR) on ogbl-citation2, and 62.42% (HR@20) on ogbl-ddi over plain GCN.

Reading between the lines

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

  • The adaptivity result (Remark 4.4) implies a transfer experiment the paper does not run: an encoder tuned on a dense graph should, via the adapter, re-weight toward path-based features when moved to a sparse graph, which would turn VSFs into a reusable cross-dataset structural-feature layer.
  • The same render-and-encode pipeline is a natural fit for node classification, graph classification, and heterogeneous or temporal graphs, where the local visual pattern carries information distinct from node attributes; the paper's own impact statement flags these extensions.
  • Since VSFs are permutation-sensitive, rendering effectively injects a data-augmentation family (layout seeds, rotations, flips, colors) into a permutation-equivariant model; a direct follow-up is to measure which of these visual augmentations regularize learning best.
  • A cheap control suggested by the partial-adaptivity result: replace the pretrained ResNet50 with a random-weight encoder and keep the adapter trainable. If accuracy holds, pretrained visual knowledge is not the source of the gain; if it collapses, natural-image pretraining is load-bearing.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes Graph Vision Network (GVN) and its efficient variant E-GVN, which render k-hop subgraphs around query links (or around nodes) as images via Graphviz/sfdp, encode them with a pretrained ResNet50 to obtain Visual Structural Features (VSFs), and fuse these into MPNNs (GCN, NCNC) for link prediction. The authors report consistent gains across seven datasets, including OGB large-scale graphs, and claim new SOTA results. The paper includes ablations on visualization scope, style, encoder, integration strategy, and a substructure-counting analysis.

Significance. If the reported gains reflect genuine visual-structure awareness rather than leakage of the target-edge label into the rendered image, the work would open a new and plausibly orthogonal direction for link prediction, with practical value from the efficient node-centered variant. The manuscript is generally well organized, releases code, and provides detailed ablations. However, as argued in the major comments, the current experimental protocol does not rule out a direct visual shortcut, and the substructure-counting evidence is consistent with the encoder reading edge presence. The contribution is therefore significant in potential but not yet established.

major comments (3)
  1. [Section 4.1, Algorithms 1 and 3, Appendix D] The central claim that VSFs encode link-relevant structure is threatened by a label-leakage mechanism. In Algorithm 1 (GVN), Step 1 extracts the k-hop subgraph S^k_uv from G and passes it to the visualizer without deleting the query edge (u,v); Section 4.1 only states that the link is 'masked for prediction,' which is a drawing operation. Since the default layout is Graphviz's force-directed sfdp, the presence of (u,v) in the graph object used for layout pulls the two highlighted endpoint nodes together, so the rendered image encodes the label through spatial proximity even when the edge stroke is not drawn. For E-GVN (Algorithm 3), the node-centered subgraph S^k_v is visualized with no masking at all, so a positive training edge (u,v) is rendered as a drawn edge inside the image of v, whereas a negative pair has no such edge; the model can therefore read the label directly from image content. Appendix D further states that on ogbl-collab validation edges are added to G at test time, making the shortcut available in validation. This leakage would explain the large gains in Table 3 (e.g., 38.86% relative improvement for E-GVNGCN on ogbl-collab) and the near-perfect substructure counts in Table 1. The authors must either remove the target edge from the subgraph before layout (for both GVN and E-GVN) or otherwise demonstrate that the layout and rendering do not convey the target edge's existence; without such a control, the empirical core of the paper does not support RQ2.
  2. [Table 1 and Section 4.2] The substructure-counting experiment is presented as evidence that VSFs confer fine-grained substructure awareness, but the near-zero normalized MSE values (e.g., 6.76E-9 for VSF+GCN) indicate that the vision encoder essentially reads the edge set directly from the rendered graph. This is not evidence of a generally useful visual structural feature; it is exactly what one would expect if the image contains the adjacency information. To support Remark 4.2, the substructure-counting task should be run with a protocol where the image is the only input and the target substructure is not trivially readable as drawn edges, or the experiment should be reinterpreted as a demonstration of edge-presence reading rather than structural generalization.
  3. [Section 5.1 and Table 3] The claim that GVN/E-GVN 'achieve new SOTA results' is stronger than the evidence supports. Baseline numbers are taken directly from Wang et al. (2024) rather than re-run under the authors' protocol; as Appendix D notes, the treatment of validation edges as message-passing paths differs across datasets and baselines. Even setting aside the leakage issue, the comparison would benefit from first-hand baseline runs and standard significance tests; at least one overlap of the reported mean and standard deviation (e.g., E-GVNNCNC vs NCNC hit@100 on Cora, 91.47±0.36 vs 89.65±1.36) is not formally assessed.
minor comments (5)
  1. [Section 3, Eq. (1)] The message-passing update equation has unbalanced parentheses: the expression should be corrected to clearly delimit the aggregation set and the update inputs.
  2. [Table 7] The header of Table 7 contains a typo: 'Igraj' should be 'Igraph'.
  3. [Section 4.4, Algorithms 2 and 3] The terms 'Adaptor' and 'Adapter' are used interchangeably; please unify the spelling throughout.
  4. [Appendix C, Table 12] The header of Table 12 lists 'Collab,PPA,Collab,DDI' with 'Collab' appearing twice; it should be 'Collab, PPA, DDI, Citation2'.
  5. [Section 5.5, Table 11] The comparison with 2-dimensional coordinates uses raw coordinates fed into a GCN, not through the vision encoder; clarify that this comparison conflates encoder capacity with information content, and consider a control that passes the same image without the queried-edge layout cue.

Circularity Check

0 steps flagged · score 2.0 of 10

No material circularity: central results are measured on external benchmarks; the only same-group citation (GITA) is motivational and not load-bearing.

full rationale

GVN and E-GVN propose a concrete pipeline: render k-hop subgraphs to images with Graphviz, extract VSFs with a pretrained ResNet50, and fuse them into MPNNs. The central claim (vision awareness improves link prediction) is tested on seven external benchmarks against baselines from other papers, not against quantities derived from the method's own outputs. No fitted parameter is relabeled as a prediction: the visual encoder is frozen or fine-tuned end-to-end on the same supervised train/test protocol as the MPNN, and the reported numbers are standard test metrics. The only same-group citation is GITA (Wei et al., 2024), used to motivate that visual perception can benefit graph reasoning; it does not supply the link-prediction results or the SOTA comparisons, so it is not load-bearing. The substructure-counting and SF-reproduction analyses are auxiliary demonstrations rather than inputs to the main claim. A possible validity threat (not a circularity) is the queried-edge masking: S^k_uv and S^k_v are defined as the subgraphs 'around' the target link, so the target edge is included in the extracted subgraph; if the layout or rendering does not remove it before feature extraction, the image could encode label information. This is a correctness/leakage concern that would need code inspection, not a derivation that reduces to its own inputs by construction. Overall, the derivation is self-contained with respect to external evidence, so the circularity score is low.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The framework introduces no new physical or mathematical entities. It relies on three hand-chosen design parameters (scope k, fusion strategy, image style) and on four assumptions about graph rendering and encoder transfer. The central empirical claim is not circular with respect to these choices, but the choices do affect the magnitude and generality of the reported gains.

free parameters (3)
  • Visualization scope k = 2 (default; tuned in {1,2,3})
    Controls how much local topology is drawn and encoded; k=2 is selected by validation and used for the reported core results.
  • Feature integration strategy = Attention-based (default)
    Three strategies are proposed in Section 4.3; attention-based integration is chosen because Table 9 shows it usually performs best.
  • Image style configuration = Graphviz with sfdp layout, brown center node or link, white fill, no labels
    Hand-chosen in Appendix F; Table 5 shows inconsistent styles degrade results, so the reported numbers depend on this style choice.
assumptions (4)
  • standard math MPNNs are limited by the 1-dimensional Weisfeiler-Lehman test and cannot distinguish some links involving isomorphic nodes.
    Admitted via Zhang et al. (2021) and Chamberlain et al. (2023); used in Section 4.2 to motivate visual structural features.
  • domain assumption The locality principle holds: k-hop subgraphs with k at most 3 contain the signal needed for link prediction.
    Invoked in Section 4.1 to justify limiting the visualization scope to a small local subgraph.
  • domain assumption A fixed force-directed 2D layout preserves link-relevant topology in pixel form.
    Assumed in Section 4.1 and Appendix F; no proof that sfdp layouts preserve the separations needed for link discrimination.
  • domain assumption ImageNet-pretrained ResNet50 features transfer to synthetic graph drawings.
    Used by default in Section 4.2 and Appendix D; supported indirectly by Tables 7 and 8, but only empirically.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Open Your Eyes: Vision Enhances Message Passing Neural Networks in Link Prediction." pith.science (2026). https://pith.science/paper/DCPROB6T

@misc{pith2026250508266,
  author       = {Pith},
  title        = {Pith review of: Open Your Eyes: Vision Enhances Message Passing Neural Networks in Link Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DCPROB6T}},
  note         = {Machine review of arXiv:2505.08266}
}
read the original abstract

Message-passing graph neural networks (MPNNs) and structural features (SFs) are cornerstones for the link prediction task. However, as a common and intuitive mode of understanding, the potential of visual perception has been overlooked in the MPNN community. For the first time, we equip MPNNs with vision structural awareness by proposing an effective framework called Graph Vision Network (GVN), along with a more efficient variant (E-GVN). Extensive empirical results demonstrate that with the proposed frameworks, GVN consistently benefits from the vision enhancement across seven link prediction datasets, including challenging large-scale graphs. Such improvements are compatible with existing state-of-the-art (SOTA) methods and GVNs achieve new SOTA results, thereby underscoring a promising novel direction for link prediction.

Figures

Figures reproduced from arXiv: 2505.08266 by the authors.

Figure 1
Figure 1. An illustration of the GVN framework (left) and its efficient variant E-GVN (middle), with the example illustrations of subgraph visualization images (right). GVN and E-GVN seamlessly incorporate visual awareness of graph structures into MPNNs, which boosts the link prediction capabilities while maintaining the orthogonal compatibility with existing methods. where h t v is the representation of node v after t layers… view at source ↗
Figure 2
Figure 2. Illustrations of visual graph images with different k’s. Decoupled Vision Scope. While the perception scope ex￾3 [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. (a) An illustration of the challenge in distinguishing links with isomorphic nodes. (b) and (c) are the 1-hop subgraphs sur￾rounding the links (v1, v2) and (v1, v3), respectively. scope, our method allows the model to set the independent scope of MPNNs via its depth. This enables the model to benefit both from the refined structure awareness pro￾vided by VSFs and the capability to access more structure information i… view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Reproduction ratios of various structural features based on VSFs on the ogbl-ddi dataset. of various SFs based on VSFs on ogbl-ddi, by feeding the VSFs extracted from the pretrained ResNet50 into a train￾able 3-layer MLP. As can be seen, VSFs effectively capture most o…
Figure 5
Figure 5. Figure 5: compares the inference time and GPU memory for inferring one batch of samples from Cora (including pre-processing time). Among all the methods in compari￾son, GVN is the most time-consuming, followed by SEAL and NBFnet. These three methods also require consider￾ably mo…
Figure 6
Figure 6. Figure 6: Node-centered examples of the visualized images by various visualizers (a) Graphviz (b) Matplotlib (c) Igraph. G. Style Consistency Ablation Experimental Details In this section, we provide a detailed description of the experimental setup used in [PITH_FULL_IMAGE:figu…
Figure 7
Figure 7. Figure 7: Link-centered subgraph visualization with “No-label” labeling scheme [PITH_FULL_IMAGE:figures/full_fig_p019_7.png]
Figure 8
Figure 8. Figure 8: Link-centered subgraph visualization with “Re-label” labeling scheme [PITH_FULL_IMAGE:figures/full_fig_p019_8.png]
Figure 9
Figure 9. Figure 9: Link-centered subgraph visualization with “Unique” labeling scheme. 19 [PITH_FULL_IMAGE:figures/full_fig_p019_9.png]
Figure 10
Figure 10. Figure 10: Node-centered subgraph visualization with “No-label” labeling scheme [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]
Figure 11
Figure 11. Figure 11: Node-centered subgraph visualization with “Re-label” labeling scheme. 20 [PITH_FULL_IMAGE:figures/full_fig_p020_11.png]
Figure 12
Figure 12. Figure 12: Node-centered subgraph visualization with “Unique” labeling scheme. 21 [PITH_FULL_IMAGE:figures/full_fig_p021_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 51 canonical work pages

  1. [1]

    Adamic, L. A. and Adar, E. Friends and neighbors on the web. Social Networks, 25 0 (3): 0 211--230, 2003

    work page 2003

  2. [2]

    L., and Parikh, D

    Antol, S., Agrawal, A., Lu, J., Mitchell, M., Batra, D., Zitnick, C. L., and Parikh, D. VQA : Visual question answering. In IEEE International Conference on Computer Vision, pp.\ 2425--2433, 2015

    work page 2015

  3. [3]

    u ck, F., K \

    Arvind, V., Fuhlbr \"u ck, F., K \"o bler, J., and Verbitsky, O. On weisfeiler-leman invariance: Subgraph counts and related graph properties. Journal of Computer and System Sciences, 113: 0 42--59, 2020

    work page 2020

  4. [4]

    and Albert, R

    Barab \'a si, A.-L. and Albert, R. Emergence of scaling in random networks. Science, 286 0 (5439): 0 509--512, 1999

    1999

  5. [5]

    Translating embeddings for modeling multi-relational data

    Bordes, A., Usunier, N., Garcia-Duran, A., Weston, J., and Yakhnenko, O. Translating embeddings for modeling multi-relational data. Advances in Neural Information Processing Systems, 26, 2013

    work page 2013

  6. [6]

    M., Bruna, J., LeCun, Y., Szlam, A., and Vandergheynst, P

    Bronstein, M. M., Bruna, J., LeCun, Y., Szlam, A., and Vandergheynst, P. Geometric deep learning: Going beyond E uclidean data. IEEE Signal Processing Magazine, 34 0 (4): 0 18--42, 2017

    work page 2017

  7. [7]

    P., Shirobokov, S., Rossi, E., Frasca, F., Markovich, T., Hammerla, N

    Chamberlain, B. P., Shirobokov, S., Rossi, E., Frasca, F., Markovich, T., Hammerla, N. Y., Bronstein, M. M., and Hansmire, M. Graph neural networks for link prediction with subgraph sketching. In International Conference on Learning Representations, 2023

    work page 2023

  8. [8]

    Structure-aware transformer for graph representation learning

    Chen, D., O’Bray, L., and Borgwardt, K. Structure-aware transformer for graph representation learning. In International Conference on Machine Learning, pp.\ 3469--3489, 2022

    work page 2022

Show all 59 references
  1. [9]

    and Wei, Y

    Chen, K. and Wei, Y. UPCS : Unbiased persona construction for dialogue generation. In International Conference on Acoustics, Speech and Signal Processing, pp.\ 1--5, 2025

    2025

  2. [10]

    Can graph neural networks count substructures? Advances in Neural Information Processing Systems, 33: 0 10383--10395, 2020

    Chen, Z., Chen, L., Villar, S., and Bruna, J. Can graph neural networks count substructures? Advances in Neural Information Processing Systems, 33: 0 10383--10395, 2020

    2020

  3. [11]

    Cluster- GCN : An efficient algorithm for training deep and large graph convolutional networks

    Chiang, W.-L., Liu, X., Si, S., Li, Y., Bengio, S., and Hsieh, C.-J. Cluster- GCN : An efficient algorithm for training deep and large graph convolutional networks. In International Conference on Knowledge Discovery & Data Mining, pp.\ 257--266, 2019

    2019

  4. [12]

    Which modality should I use -- T ext, motif, or image?: Understanding graphs with large language models

    Das, D., Gupta, I., Srivastava, J., and Kang, D. Which modality should I use -- T ext, motif, or image?: Understanding graphs with large language models. Preprint arXiv:2311.09862, 2023

    2023 arXiv

  5. [13]

    Dijkstra, E. W. A note on two problems in connexion with graphs. Numerische Mathematik, 1959

    1959

  6. [14]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weis, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. An image is worth 16x16 words: Transformers for image recognition at scale. In International Conference on Learning Representations, 2021

    2021

  7. [15]

    Gabor Csardi, T. N. The igraph software package for complex network research. InterJournal, Complex Systems: 0 1695, 2006

    2006

  8. [16]

    Gansner, E. R. and North, S. C. An open graph visualization system and its applications to software engineering. Software: Practice and Experience, 30 0 (11): 0 1203--1233, 2000

    2000

  9. [17]

    and Leskovec, J

    Grover, A. and Leskovec, J. node2vec: Scalable feature learning for networks. In International Conference on Knowledge Discovery and Data Mining, pp.\ 855--864, 2016

    2016

  10. [18]

    Inductive representation learning on large graphs

    Hamilton, W., Ying, Z., and Leskovec, J. Inductive representation learning on large graphs. Advances in Neural Information Processing Systems, 30, 2017

    2017

  11. [19]

    Deep residual learning for image recognition

    He, K., Zhang, X., Ren, S., and Sun, J. Deep residual learning for image recognition. In IEEE Conference on Computer Vision and Pattern Recognition, pp.\ 770--778, 2016

    2016

  12. [20]

    Lightgcn: Simplifying and powering graph convolution network for recommendation

    He, X., Deng, K., Wang, X., Li, Y., Zhang, Y., and Wang, M. Lightgcn: Simplifying and powering graph convolution network for recommendation. In International Conference on Research and Development in Information Retrieval, pp.\ 639--648, 2020

    2020

  13. [21]

    Open graph benchmark: Datasets for machine learning on graphs

    Hu, W., Fey, M., Zitnik, M., Dong, Y., Ren, H., Liu, B., Catasta, M., and Leskovec, J. Open graph benchmark: Datasets for machine learning on graphs. Advances in Neural Information Processing Systems, 33: 0 22118--22133, 2020

    2020

  14. [22]

    and Ribeiro, A

    Kanatsoulis, C. and Ribeiro, A. Counting graph substructures with graph neural networks. In International Conference on Learning Representations, 2024

    2024

  15. [23]

    Kenta Oono, T. S. Graph neural networks exponentially lose expressive power for node classification. In International Conference on Learning Representations, 2020

    2020

  16. [24]

    Kingma, D. P. Adam: A method for stochastic optimization. Preprint arXiv:1412.6980, 2014

    2014 arXiv

  17. [25]

    Kipf, T. N. and Welling, M. Semi-supervised classification with graph convolutional networks. In International Conference on Learning Representations, 2017

    2017

  18. [26]

    Evaluating graph neural networks for link prediction: Current pitfalls and new benchmarking

    Li, J., Shomer, H., Mao, H., Zeng, S., Ma, Y., Shah, N., Tang, J., and Yin, D. Evaluating graph neural networks for link prediction: Current pitfalls and new benchmarking. Advances in Neural Information Processing Systems, 36, 2024

    2024

  19. [27]

    Distance encoding: Design provably more powerful neural networks for graph representation learning

    Li, P., Wang, Y., Wang, H., and Leskovec, J. Distance encoding: Design provably more powerful neural networks for graph representation learning. Advances in Neural Information Processing Systems, 33: 0 4465--4478, 2020

    2020

  20. [28]

    Deeper insights into graph convolutional networks for semi-supervised learning

    Li, Q., Han, Z., and Wu, X.-M. Deeper insights into graph convolutional networks for semi-supervised learning. In AAAI Conference on Artificial Intelligence, 2018

    2018

  21. [29]

    and Kleinberg, J

    Liben-Nowell, D. and Kleinberg, J. The link prediction problem for social networks. In International Conference on Information and Knowledge Management, pp.\ 556--559, 2003

    2003

  22. [30]

    Large language models can be guided to evade ai-generated text detection

    Lu, N., Liu, S., He, R., Ong, Y.-S., Wang, Q., and Tang, K. Large language models can be guided to evade ai-generated text detection. Transactions on Machine Learning Research

  23. [31]

    Less is more: Understanding word-level textual adversarial attack via n -gram frequency descend

    Lu, N., Liu, S., Zhang, Z., Wang, Q., Liu, H., and Tang, K. Less is more: Understanding word-level textual adversarial attack via n -gram frequency descend. In IEEE Conference on Artificial Intelligence, pp.\ 823--830, 2024

    2024

  24. [32]

    Safe delta: Consistently preserving safety when fine-tuning LLM s on diverse datasets

    Lu, N., Liu, S., Wu, J., Chen, W., Zhang, Z., Ong, Y.-S., Wang, Q., and Tang, K. Safe delta: Consistently preserving safety when fine-tuning LLM s on diverse datasets. Preprint arXiv:2505.12038, 2025

    2025 arXiv

  25. [33]

    K., Nigam, K., Rennie, J., and Seymore, K

    McCallum, A. K., Nigam, K., Rennie, J., and Seymore, K. Automating the construction of internet portals with machine learning. Information Retrieval, 3: 0 127--163, 2000

    2000

  26. [34]

    L., Lenssen, J

    Morris, C., Ritzert, M., Fey, M., Hamilton, W. L., Lenssen, J. E., Rattan, G., and Grohe, M. Weisfeiler and L eman go neural: Higher-order graph neural networks. In AAAI Conference on Artificial Intelligence, pp.\ 4602--4609, 2019

    2019

  27. [35]

    Query-driven active surveying for collective classification

    Namata, G., London, B., Getoor, L., Huang, B., and Edu, U. Query-driven active surveying for collective classification. In International Workshop on Mining and Learning with Graphs, 2012

    2012

  28. [36]

    Deepwalk: Online learning of social representations

    Perozzi, B., Al-Rfou, R., and Skiena, S. Deepwalk: Online learning of social representations. In International Conference on Knowledge Discovery and Data Mining, pp.\ 701--710, 2014

    2014

  29. [37]

    Collective classification in network data

    Sen, P., Namata, G., Bilgic, M., Getoor, L., Galligher, B., and Eliassi-Rad, T. Collective classification in network data. AI Magazine, 29 0 (3): 0 93--93, 2008

    2008

  30. [38]

    Resisting over-smoothing in graph neural networks via dual-dimensional decoupling

    Shen, W., Ye, M., and Huang, W. Resisting over-smoothing in graph neural networks via dual-dimensional decoupling. In International Conference on Multimedia, pp.\ 5800--5809, 2024

    2024

  31. [39]

    and Zisserman, A

    Simonyan, K. and Zisserman, A. Very deep convolutional networks for large-scale image recognition. In 3rd International Conference on Learning Representations (ICLR 2015). Computational and Biological Learning Society, 2015

    2015

  32. [40]

    Line: Large-scale information network embedding

    Tang, J., Qu, M., Wang, M., Zhang, M., Yan, J., and Mei, Q. Line: Large-scale information network embedding. In International Conference on World Wide Web, pp.\ 1067--1077, 2015

    2015

  33. [41]

    Matplotlib for Python Developers

    Tosi, S. Matplotlib for Python Developers. Packt Publishing Ltd, 2009

    2009

  34. [42]

    Graph attention networks

    Velickovic, P., Cucurull, G., Casanova, A., Romero, A., Lio, P., Bengio, Y., et al. Graph attention networks. In International Conference on Learning Representations, 2018

    2018

  35. [43]

    Neural common neighbor with completion for link prediction

    Wang, X., Yang, H., and Zhang, M. Neural common neighbor with completion for link prediction. In International Conference on Learning Representations, 2024

    2024

  36. [44]

    KICGPT : Large language model with knowledge in context for knowledge graph completion

    Wei, Y., Huang, Q., Zhang, Y., and Kwok, J. KICGPT : Large language model with knowledge in context for knowledge graph completion. In Findings of the Association for Computational Linguistics, pp.\ 8667--8683, 2023

    2023

  37. [45]

    Gita: Graph to visual and textual integration for vision-language graph reasoning

    Wei, Y., Fu, S., Jiang, W., Zhang, Z., Zeng, Z., Wu, Q., Kwok, J., and Zhang, Y. Gita: Graph to visual and textual integration for vision-language graph reasoning. Advances in Neural Information Processing Systems, 37: 0 44--72, 2024

    2024

  38. [46]

    Wu, Z., Pan, S., Chen, F., Long, G., Zhang, C., and Yu, P. S. A comprehensive survey on graph neural networks. IEEE Transactions on Neural Networks and Learning Systems, 2020

    2020

  39. [47]

    How powerful are graph neural networks? Technical report, 2019

    Xu, K., Hu, W., Leskovec, J., and Jegelka, S. How powerful are graph neural networks? Technical report, 2019

    2019

  40. [48]

    Prediction of drug--target interaction networks from the integration of chemical and genomic spaces

    Yamanishi, Y., Araki, M., Gutteridge, A., Honda, W., and Kanehisa, M. Prediction of drug--target interaction networks from the integration of chemical and genomic spaces. Bioinformatics, 24 0 (13): 0 i232--i240, 2008

    2008

  41. [49]

    An efficient subgraph GNN with provable substructure counting power

    Yan, Z., Zhou, J., Gao, L., Tang, Z., and Zhang, M. An efficient subgraph GNN with provable substructure counting power. In ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pp.\ 3702--3713, 2024

    2024

  42. [50]

    M., Ying, R., and Leskovec, J

    You, J., Gomes-Selman, J. M., Ying, R., and Leskovec, J. Identity-aware graph neural networks. In AAAI Conference on Artificial Intelligence, pp.\ 10737--10745, 2021

    2021

  43. [51]

    Yun, S., Kim, S., Lee, J., Kang, J., and Kim, H. J. Neo- GNN s: Neighborhood overlap-aware graph neural networks for link prediction. Advances in Neural Information Processing Systems, 34: 0 13683--13694, 2021

    2021

  44. [52]

    Graph SAINT : Graph sampling based inductive learning method

    Zeng, H., Zhou, H., Srivastava, A., Kannan, R., and Prasanna, V. Graph SAINT : Graph sampling based inductive learning method. Preprint arXiv:1907.04931, 2019

    1907 arXiv

  45. [53]

    Decoupling the depth and scope of graph neural networks

    Zeng, H., Zhang, M., Xia, Y., Srivastava, A., Malevich, A., Kannan, R., Prasanna, V., Jin, L., and Chen, R. Decoupling the depth and scope of graph neural networks. Advances in Neural Information Processing Systems, 34: 0 19665--19679, 2021

    2021

  46. [54]

    Vision-language models for vision tasks: A survey

    Zhang, J., Huang, J., Jin, S., and Lu, S. Vision-language models for vision tasks: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    2024

  47. [55]

    and Chen, Y

    Zhang, M. and Chen, Y. Link prediction based on graph neural networks. Advances in Neural Information Processing Systems, 31, 2018

    2018

  48. [56]

    Labeling trick: A theory of using graph neural networks for multi-node representation learning

    Zhang, M., Li, P., Xia, Y., Wang, K., and Jin, L. Labeling trick: A theory of using graph neural networks for multi-node representation learning. Advances in Neural Information Processing Systems, 34: 0 9061--9073, 2021

    2021

  49. [57]

    Predicting missing links via local information

    Zhou, T., L \"u , L., and Zhang, Y.-C. Predicting missing links via local information. The European Physical Journal B, 71: 0 623--630, 2009

    2009

  50. [58]

    Neural bellman-ford networks: A general graph neural network framework for link prediction

    Zhu, Z., Zhang, Z., Xhonneux, L.-P., and Tang, J. Neural bellman-ford networks: A general graph neural network framework for link prediction. Advances in Neural Information Processing Systems, 34: 0 29476--29490, 2021

    2021

  51. [59]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.