Pith. sign in

REVIEW 3 major objections 4 minor 53 references

Data-Driven Self-Supervised Graph Representation Learning

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · 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 self-supervised graph representation learning can replace hand-crafted augmentations with augmentations learned from the graph's own signal, and reports matching or beating state-of-the-art baselines on node…

desk verdict Solid empirical graph SSL paper with a genuinely new learned-augmentation design, but the experiments don't actually prove that the learned augmentations are what drive the gains. read the letter →

arxiv 2412.18316 v1 pith:AWXHEFVM submitted 2024-12-24 cs.LG cs.AI

Ahmed E. Samy , Zekarias T. Kefatoa , Sarunas Girdzijauskasa This is my paper · ORCID
classification cs.LGcs.AI
keywords self-supervisedlearninggraphrepresentationdataaugmentationlearnablecontrastivenodeclassificationpropertypredictionheterogeneousgraphs
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 argues that the main bottleneck in self-supervised graph representation learning is choosing good graph augmentations, and that hand-picked heuristics often destroy task-relevant structure. It proposes DSGRL, a method that learns the augmentation from the graph's own feature and topological signals, jointly with the representation. A feed-forward network learns multi-view feature perturbations, and a GNN-based module learns a high-order similarity graph by keeping edges between nodes whose learned vectors agree strongly. Tested on nine node-classification datasets and eight graph-property datasets, the authors report that the method matches or outperforms state-of-the-art self-supervised baselines and is comparable to semi-supervised methods. If so, self-supervised graph learning would no longer need dataset-specific hand-designed augmentations.

What carries the argument

The argument is carried by three components. The feature augmenter is a pair of feed-forward networks $f_{\Theta_1}$ and $f_{\Theta_2}$ that map the node feature matrix $X$ into two views $X_1, X_2$. The topology augmenter is a GNN $h_\Phi$ that computes high-order (multi-hop) node embeddings $H$ from the graph, then applies the threshold rule $a'_{ij} = h_i^\top h_j$ if $h_i^\top h_j > \mathbb{E}_k[h_i^\top h_k]$, otherwise $0$, producing a weighted high-order adjacency matrix $A'$. A shared GNN encoder $h_\Theta$ maps the two views to latent representations $Z_1, Z_2$, and the training loss combines an invariance term $\|Z_1 - Z_2\|_F$ with regularizers that keep latent dimensions spread out and decorrelated, plus an orthogonality penalty on the augmenter weights so the two feature augmenters do not collapse into one another. Because all these parameters are optimized together, the augmentation is shaped by the graph's own signal rather than chosen from a fixed set of perturbations.

What would settle it

Train DSGRL on a labeled dataset, freeze the learned topology augmenter, and replace its weighted edges with a random graph drawn uniformly among node pairs at the same edge density. If random-topology DSGRL matches learned-topology DSGRL on the downstream linear-evaluation task, then the claim that the learned augmentation itself carries the task-relevant signal is falsified; if the learned graph does better, the claim is supported.

Watch

Extended reading notes

Core claim

The central claim is that augmentation for graph self-supervision can be learned, not hand-picked. DSGRL builds two complementary views: feature views produced by two small feed-forward networks applied to the node feature matrix, and a topological view produced by a GNN that embeds nodes and then keeps a weighted edge between two nodes only when their dot-product similarity exceeds the expected similarity to the other nodes. The two views are encoded by a shared GNN, and the whole system is optimized end-to-end with a loss that asks the two representations to agree, keeps their dimensions spread out and mutually independent, and forces the two feature augmenters to be different from each other. The paper reports that this jointly learned augmentation matches or surpasses state-of-the-art self-supervised baselines on homogeneous and heterogeneous node classification and on graph property prediction, and that it remains trainable on large graphs where full-batch contrastive baselines run out of GPU memory.

Load-bearing premise

The load-bearing premise is that edges between nodes whose learned high-order feature dot products exceed the expected similarity form a view that preserves the information needed for the downstream task; the paper provides no proof that this thresholded similarity graph keeps task-relevant semantics, and it notes the topology augmenter performs poorly on chemical datasets.

Editorial extensions

If this is right

  • One data-driven pipeline can replace dataset-specific augmentation heuristics across homogeneous, heterogeneous, chemical, and social graphs.
  • The feature and topology augmenters are complementary and can be combined, so a single method can adapt its view generation to the graph type it encounters.
  • On large graphs such as Yelp and Reddit the method trains where full-batch contrastive baselines exhaust GPU memory, because it avoids negative sampling and large-batch requirements.
  • The reported node-classification accuracy is close to the semi-supervised references, which suggests label-free pretraining can substitute for much of the labeled training data.
  • The paper's own untrained random variants are strong competitors, especially on chemical datasets, so the marginal benefit of the learned augmentation should be judged against those random baselines, not only against older heuristics.

Reading between the lines

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

  • Editorial inference: a dataset-dependent selector that picks between the feature and topology augmenters would probably improve average performance, since the paper reports topology augmentation underperforms on chemical datasets but does not propose such a selector.
  • Editorial inference: the threshold rule, which compares each node-pair similarity to the expected similarity to all other nodes, is one natural choice among many; testing cosine similarity or a learned metric would show how much the result depends on that specific function.
  • Editorial inference: because DSGRL requires no meta-paths, the single heterogeneous-graph experiment suggests the approach could carry over to knowledge graphs and other multi-relational networks, but that extension is not yet demonstrated.
  • Editorial inference: the strength of the untrained variants implies that future augmentation methods should be benchmarked against random augmentations of equal strength; otherwise the encoder's inductive bias can be mistaken for an augmentation gain.
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 / 4 minor

Summary. The paper proposes DSGRL, a self-supervised graph representation learning method with learnable feature and topology augmentations. Feature augmentation is implemented by two feed-forward networks that produce augmented feature views, while topology augmentation is implemented by a GNN that computes high-order node similarities, which are thresholded to form a high-order adjacency view. The augmenters are trained jointly with a shared GNN encoder using a VICReg-style invariance, variance, and covariance objective with an additional model-orthogonality regularizer. The method is evaluated on node classification (including heterogeneous graphs) and graph property prediction, comparing against self-supervised baselines and reporting untrained variants Random-F and Random-T. The central claim is that DSGRL automatically learns augmentations from the graph signal, avoiding hand-designed heuristics, while matching or outperforming existing self-supervised baselines.

Significance. If the central claim holds, DSGRL would provide a generally applicable, data-driven alternative to heuristic graph augmentations, with the attraction of being usable for both homogeneous and heterogeneous graphs. The paper is empirically extensive: it uses multiple public benchmarks, follows standard linear-evaluation protocols, tunes baselines, includes untrained controls, and points to source code and an appendix. These are strengths. The main qualification is that the evidence for the claim that learned augmentations drive the gains is incomplete, because the untrained variants are strong competitors on several datasets and no control isolates the effect of augmenter training from encoder training. The conceptual contribution is therefore plausible but needs a sharper experiment before the strongest conclusions are justified.

major comments (3)
  1. [Section 2.5, Tables 3 and 6] This is the load-bearing weakness for the paper's headline contribution. Please add the frozen-augmenter ablation or substantially weaken the causal interpretation.
  2. [Section 2.3.2 and Section 5] The topology augmenter is only partially learned. The similarity function g(hi,hj)=hi^T hj and the threshold Ek[g(hi,hk)] are fixed modeling choices, and the paper provides no analysis showing that the resulting thresholded high-order network preserves the task-relevant information needed for downstream classification. The paper itself notes that topological augmentation performs poorly on chemical datasets. This does not invalidate the empirical results, but it weakens the claim that the augmentation is 'suitable' and learned from the signal; an ablation varying the similarity function or threshold, or a theoretical justification of the thresholding operation, would make the data-driven claim more convincing.
  3. [Tables 4 and 5] The scalability and heterogeneous-graph claims are based on results with no standard deviations or significance tests: Table 4 reports single numbers for Yelp and Reddit, and Table 5 reports point estimates for a single split on IMDB. Given the paper's emphasis on scalability and generality, these claims should be supported with repeated runs or explicit acknowledgment of the limited evidence.
minor comments (4)
  1. [Section 3.1 and Conclusion] The dataset count is inconsistent: Section 3.1 says the NC datasets are 8, while the abstract and conclusion say nine node classification datasets. Please reconcile this count by explicitly counting the heterogeneous IMDB dataset.
  2. [Table 6] There is a typo in the table header: 'Ramdom-T' should be 'Random-T'.
  3. [Section 5] The conclusion states that 'DSGRL surpasses the baseline SOTA approaches,' but in several cases the differences are within one standard deviation or tied (e.g., NCI1 in Table 6 where Feature reports 75.0 vs. GraphCL's 75.2). A more cautious wording such as 'matches or outperforms' would be consistent with the abstract and the displayed results.
  4. [Section 2.3.2] The notation A' = t_Phi'(A) = g(H,H) is slightly confusing because g is described as a function on pairs of node vectors, while the matrix A' has entries a'_ij. Please clarify that g is applied elementwise to pairs of rows of H.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: DSGRL's central derivation is self-contained and its benchmark evaluation is held out.

full rationale

The paper's core claims are empirical and its training objective is defined independently of the downstream evaluation. The augmentation models (f_Theta1, f_Theta2 for features; t'_Phi for topology) are parameterized functions optimized with a VICReg-style loss (Eqs. 1-6); no quantity appearing in the loss is a fitted proxy for the reported test accuracies. The evaluation uses public splits and held-out test sets, and hyperparameters are tuned on validation splits, so the reported gains are not forced by construction. The topology augmenter's similarity threshold (Section 2.3.2) is a stated heuristic modeling choice, not a circular definition of the result. Self-citations [17, 18, 29] are contextual (examples of GRL methods and of extending BGRL to graphs) and are not load-bearing for the central derivation; the principled collapse-prevention formulation is explicitly attributed to the external VICReg work [1]. The paper's own admission that untrained Random-F/Random-T variants are strong competitors (Tables 3 and 6, Section 3.5) is an honest limitation about isolating the contribution of learned augmentations, but that is an experimental-design concern rather than a circular reduction. No equation or fitted parameter is renamed as a prediction, and no uniqueness claim is imported from prior work by the same authors. Therefore, no specific circular step can be exhibited.

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

The method rests on standard SSL assumptions and one paper-specific heuristic: thresholded similarity-based topology augmentation. Hyperparameters are tuned via Bayesian optimization, making them fitted values rather than derivations. No new physical or conceptual entities are introduced.

free parameters (3)
  • hyperparameter weights alpha, beta, gamma, lambda = set to 1 in most cases, tuned via Optuna
    Control trade-off between invariance, variance, covariance, and model regularization in Eq. 6; tuned per dataset.
  • augmentation dimension D1 = not reported
    Dimension of learned feature augmentations; part of hyperparameter search.
  • latent dimension D = 64 for DSGRL, 128 for baselines
    Representation dimension; chosen for the model and to justify concatenation of views.
assumptions (4)
  • domain assumption VICReg-style variance-invariance-covariance regularization yields useful representations
    Paper follows [1] (Bardes et al.) without proof; this is the learning objective.
  • ad hoc to paper High-order node similarities from a GNN embedding capture useful topological signal
    Section 2.3.2 assumes that hi·hj > expected similarity identifies semantically meaningful edges; no theoretical justification.
  • standard math Linear evaluation on test splits measures representation quality
    Uses standard linear probe protocol from [31].
  • domain assumption Message-passing GNN is a suitable encoder for both homogeneous and heterogeneous graphs
    Uses GCN-style encoder throughout.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Data-Driven Self-Supervised Graph Representation Learning." pith.science (2026). https://pith.science/paper/AWXHEFVM

@misc{pith2026241218316,
  author       = {Pith},
  title        = {Pith review of: Data-Driven Self-Supervised Graph Representation Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AWXHEFVM}},
  note         = {Machine review of arXiv:2412.18316}
}
read the original abstract

Self-supervised graph representation learning (SSGRL) is a representation learning paradigm used to reduce or avoid manual labeling. An essential part of SSGRL is graph data augmentation. Existing methods usually rely on heuristics commonly identified through trial and error and are effective only within some application domains. Also, it is not clear why one heuristic is better than another. Moreover, recent studies have argued against some techniques (e.g., dropout: that can change the properties of molecular graphs or destroy relevant signals for graph-based document classification tasks). In this study, we propose a novel data-driven SSGRL approach that automatically learns a suitable graph augmentation from the signal encoded in the graph (i.e., the nodes' predictive feature and topological information). We propose two complementary approaches that produce learnable feature and topological augmentations. The former learns multi-view augmentation of node features, and the latter learns a high-order view of the topology. Moreover, the augmentations are jointly learned with the representation. Our approach is general that it can be applied to homogeneous and heterogeneous graphs. We perform extensive experiments on node classification (using nine homogeneous and heterogeneous datasets) and graph property prediction (using another eight datasets). The results show that the proposed method matches or outperforms the SOTA SSGRL baselines and performs similarly to semi-supervised methods. The anonymised source code is available at https://github.com/AhmedESamy/dsgrl/

Figures

Figures reproduced from arXiv: 2412.18316 by the authors.

Figure 1
Figure 1. The Architecture of DSGRL . The left figure (a) is based on learnable feature augmentation, where we only augment node features, X. Two learnable augmenters fΘ1 (X) and fΘ2 (X) are applied on X to obtain the learned augmentations X1 and X2, respectively. The middle figure (b) is used for learning topological augmentation. Two augmentations, an identical, tΦ(A), and learned ones, tΦ′ (A), are respectively applied on … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 41 canonical work pages

  1. [1]

    Vicreg: V ariance-invariance- covariance regularization for self-supervised learning, 2021

    Adrien Bardes, Jean Ponce, and Y ann LeCun. Vicreg: V ariance-invariance- covariance regularization for self-supervised learning, 2021

    work page 2021

  2. [2]

    Mikhail Belkin and Partha Niyogi, ‘Laplacian eigenmaps for dimensionality reduction and data representation’,Neural Computation, 15(6), 1373–1396, (2003)

    work page 2003

  3. [3]

    Piotr Bielak, Tomasz Kajdanowicz, and Nitesh V . Chawla. Graph barlow twins: A self-supervised representation learning framework for graphs, 2021

    work page 2021

  4. [4]

    Mathilde Caron, Ishan Misra, Julien Mairal, Priya Goyal, Piotr Bojanowski, and Armand Joulin, ‘Unsupervised learning of visual features by contrasting cluster assignments’,CoRR, abs/2006.09882, (2020)

    arXiv 2020

  5. [5]

    Mathilde Caron, Hugo Touvron, Ishan Misra, Hervé Jégou, Julien Mairal, Piotr Bojanowski, and Armand Joulin, ‘Emerging properties in self-supervised vision transformers’,CoRR, abs/2104.14294, (2021)

    arXiv 2021

  6. [6]

    Xinlei Chen and Kaiming He, ‘Exploring simple siamese representation learning’,CoRR, abs/2011.10566, (2020)

    arXiv 2020

  7. [7]

    Bert: Pre-training of deep bidirectional transformers for language understanding, 2019

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding, 2019

    2019

  8. [8]

    Chawla, and Ananthram Swami, ‘Metapath2vec: Scalable representation learning for heterogeneous networks’, inProceedings of the 23rd ACM SIGKDD, KDD ’17, p

    Y uxiao Dong, Nitesh V . Chawla, and Ananthram Swami, ‘Metapath2vec: Scalable representation learning for heterogeneous networks’, inProceedings of the 23rd ACM SIGKDD, KDD ’17, p. 135–144, New Y ork, NY , USA, (2017). Association for Computing Machinery

    work page 2017

Show all 53 references
  1. [9]

    A fair comparison of graph neural networks for graph classification, 2020

    Federico Errica, Marco Podda, Davide Bacciu, and Alessio Micheli. A fair comparison of graph neural networks for graph classification, 2020

    2020

  2. [10]

    ACM, (apr 2020)

    Xinyu Fu, Jiani Zhang, Ziqiao Meng, and Irwin King, ‘MAGNN: Metapath aggregated graph neural network for heterogeneous graph embedding’, in Proceedings of The W eb Conference 2020. ACM, (apr 2020)

    2020

  3. [11]

    Richemond, Elena Buchatskaya, Carl Doersch, Bernardo Avila Pires, Zhaohan Daniel Guo, Mohammad Gheshlaghi Azar, Bilal Piot, Koray Kavukcuoglu, Rémi Munos, and Michal V alko

    Jean-Bastien Grill, Florian Strub, Florent Altché, Corentin Tallec, Pierre H. Richemond, Elena Buchatskaya, Carl Doersch, Bernardo Avila Pires, Zhaohan Daniel Guo, Mohammad Gheshlaghi Azar, Bilal Piot, Koray Kavukcuoglu, Rémi Munos, and Michal V alko. Bootstrap your own latent...

    2020

  4. [12]

    node2vec: Scalable feature learning for networks, 2016

    Aditya Grover and Jure Leskovec. node2vec: Scalable feature learning for networks, 2016

    2016

  5. [13]

    Hamilton, Rex Ying, and Jure Leskovec

    William L. Hamilton, Rex Ying, and Jure Leskovec. Inductive representation learning on large graphs, 2018

    2018

  6. [14]

    Kaveh Hassani and Amir Hosein Khas Ahmadi, ‘Contrastive multi-view representation learning on graphs’,CoRR, abs/2006.05582, (2020)

    2020 arXiv

  7. [15]

    Girshick, ‘Momentum contrast for unsupervised visual representation learning’, (2019)

    Kaiming He, Haoqi Fan, Y uxin Wu, Saining Xie, and Ross B. Girshick, ‘Momentum contrast for unsupervised visual representation learning’, (2019)

    2019

  8. [16]

    2414–2424, (2021)

    Baoyu Jing, Chanyoung Park, and Hanghang Tong, ‘Hdmi: High-order deep multiplex infomax’, inProceedings of the W eb Conference 2021, pp. 2414–2424, (2021)

    2021

  9. [17]

    Kefato and Sarunas Girdzijauskas, ‘Gossip and attend: Context-sensitive graph representation learning’, in14-th International Conference on W eb and Social Media, ICWSM’20, (2020)

    Zekarias T. Kefato and Sarunas Girdzijauskas, ‘Gossip and attend: Context-sensitive graph representation learning’, in14-th International Conference on W eb and Social Media, ICWSM’20, (2020)

    2020

  10. [18]

    Kefato and Sarunas Girdzijauskas, ‘Self-supervised graph neural networks without explicit negative sampling’,CoRR, (2021)

    Zekarias T. Kefato and Sarunas Girdzijauskas, ‘Self-supervised graph neural networks without explicit negative sampling’,CoRR, (2021)

    2021

  11. [19]

    Kipf and Max Welling

    Thomas N. Kipf and Max Welling. Semi-supervised classification with graph convolutional networks, 2017

    2017

  12. [20]

    Diffusion improves graph learning, 2019

    Johannes Klicpera, Stefan Weißenberger, and Stephan Günnemann. Diffusion improves graph learning, 2019

    2019

  13. [21]

    Albert: A lite bert for self-supervised learning of language representations, 2020

    Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, and Radu Soricut. Albert: A lite bert for self-supervised learning of language representations, 2020

    2020

  14. [22]

    Qi Li, Wenping Chen, Zhaoxi Fang, Changtian Ying, and Chen Wang, ‘A multi-view contrastive learning for heterogeneous network embedding’, Scientific Reports, 13(1), 6732, (2023)

    2023

  15. [23]

    Roberta: A robustly optimized bert pretraining approach, 2019

    Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and V eselin Stoyanov. Roberta: A robustly optimized bert pretraining approach, 2019

    2019

  16. [24]

    Christopher Morris, Nils M. Kriege, Franka Bause, Kristian Kersting, Petra Mutzel, and Marion Neumann, ‘Tudataset: A collection of benchmark datasets for learning with graphs’, in ICML 2020 W orkshop on Graph Representation Learning and Beyond (GRL+ 2020), (2020)

    2020

  17. [25]

    5371–5378, (2020)

    Chanyoung Park, Donghyun Kim, Jiawei Han, and Hwanjo Y u, ‘Unsuper- vised attributed multiplex network embedding’, inProceedings of the AAAI Conference on Artificial Intelligence, volume 34, pp. 5371–5378, (2020)

    2020

  18. [26]

    701–710, New Y ork, NY , USA, (2014)

    Bryan Perozzi, Rami Al-Rfou, and Steven Skiena, ‘Deepwalk: Online learning of social representations’, inProceedings of the 20th ACM SIGKDD, KDD’14, p. 701–710, New Y ork, NY , USA, (2014). ACM

    2014

  19. [27]

    Y uxiang Ren, Bo Liu, Chao Huang, Peng Dai, Liefeng Bo, and Jiawei Zhang, ‘Heterogeneous deep graph infomax’,arXiv preprint arXiv:1911.08538, (2019)

    2019 arXiv

  20. [28]

    Multi-scale attributed node embedding, 2021

    Benedek Rozemberczki, Carl Allen, and Rik Sarkar. Multi-scale attributed node embedding, 2021

    2021

  21. [29]

    1157–1166, (2022)

    Ahmed E Samy, Lodovico Giaretta, Zekarias T Kefato, and Šar¯unas Girdz- ijauskas, ‘Schemawalk: Schema aware random walks for heterogeneous graph embedding’, inCompanion Proceedings of the W eb Conference 2022, pp. 1157–1166, (2022)

    2022

  22. [30]

    Kaplan, Jiawei Han, and Jian Peng

    Jingbo Shang, Meng Qu, Jialu Liu, Lance M. Kaplan, Jiawei Han, and Jian Peng. Meta-path guided embedding for similarity search in large-scale heterogeneous information networks, 2016

    2016

  23. [31]

    Pitfalls of graph neural network evaluation, 2019

    Oleksandr Shchur, Maximilian Mumme, Aleksandar Bojchevski, and Stephan Günnemann. Pitfalls of graph neural network evaluation, 2019

    2019

  24. [32]

    Chuan Shi, Binbin Hu, Wayne Xin Zhao, and Philip S. Y u. Heterogeneous information network embedding for recommendation, 2017

    2017

  25. [33]

    Fan-Y un Sun, Jordan Hoffman, Vikas V erma, and Jian Tang, ‘Infograph: Unsupervised and semi-supervised graph-level representation learning via mutual information maximization’, inInternational Conference on Learning Representations, (2020)

    2020

  26. [34]

    Mocl: Contrastive learning on molecular graphs with multi-level domain knowledge, 2021

    Mengying Sun, Jing Xing, Huijun Wang, Bin Chen, and Jiayu Zhou. Mocl: Contrastive learning on molecular graphs with multi-level domain knowledge, 2021

    2021

  27. [35]

    Adversarial graph augmentation to improve graph contrastive learning, 2021

    Susheel Suresh, Pan Li, Cong Hao, and Jennifer Neville. Adversarial graph augmentation to improve graph contrastive learning, 2021

    2021

  28. [36]

    Qiaoyu Tan, Sirui Ding, Ninghao Liu, Soo-Hyun Choi, Li Li, Rui Chen, and Xia Hu, ‘Graph contrastive learning with model perturbation’, (2023)

    2023

  29. [37]

    1067–1077, Republic and Canton of Geneva, CHE, (2015)

    Jian Tang, Meng Qu, Mingzhe Wang, Ming Zhang, Jun Y an, and Qiaozhu Mei, ‘Line: Large-scale information network embedding’, WWW’15, p. 1067–1077, Republic and Canton of Geneva, CHE, (2015). ACM

    2015

  30. [38]

    Bootstrapped representation learning on graphs, 2021

    Shantanu Thakoor, Corentin Tallec, Mohammad Gheshlaghi Azar, Rémi Munos, Petar V eliˇckovi´c, and Michal V alko. Bootstrapped representation learning on graphs, 2021

    2021

  31. [39]

    Augmentations in graph contrastive learning: Current methodological flaws; towards better practices, 2021

    Puja Trivedi, Ekdeep Singh Lubana, Y ujun Y an, Y aoqing Y ang, and Danai Koutra. Augmentations in graph contrastive learning: Current methodological flaws; towards better practices, 2021

    2021

  32. [40]

    Hamilton, Pietro Liò, Y oshua Bengio, and R Devon Hjelm

    Petar V eliˇckovi´c, William Fedus, William L. Hamilton, Pietro Liò, Y oshua Bengio, and R Devon Hjelm. Deep graph infomax, 2018

    2018

  33. [41]

    1631–1639, (2022)

    Can Wang, Sheng Zhou, Kang Y u, Defang Chen, Bolang Li, Y an Feng, and Chun Chen, ‘Collaborative knowledge distillation for heterogeneous information network embedding’, in Proceedings of the ACM W eb Conference 2022, pp. 1631–1639, (2022)

    2022

  34. [42]

    2022–2032, New Y ork, NY , USA, (2019)

    Xiao Wang, Houye Ji, Chuan Shi, Bai Wang, Y anfang Y e, Peng Cui, and Philip S Y u, ‘Heterogeneous graph attention network’, inThe W orld Wide W eb Conference, WWW ’19, p. 2022–2032, New Y ork, NY , USA, (2019). Association for Computing Machinery

    2019

  35. [43]

    Self-supervised heterogeneous graph neural network with co-contrastive learning, 2021

    Xiao Wang, Nian Liu, Hui Han, and Chuan Shi. Self-supervised heterogeneous graph neural network with co-contrastive learning, 2021

    2021

  36. [44]

    1726–1736, (2021)

    Xiao Wang, Nian Liu, Hui Han, and Chuan Shi, ‘Self-supervised heteroge- neous graph neural network with co-contrastive learning’, inProceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining, pp. 1726–1736, (2021)

    2021

  37. [45]

    1070–1079, (2022)

    Jun Xia, Lirong Wu, Jintao Chen, Bozhen Hu, and Stan Z Li, ‘Simgrace: A simple framework for graph contrastive learning without data augmentation’, in Proceedings of the ACM W eb Conference 2022, pp. 1070–1079, (2022)

    2022

  38. [46]

    Self-supervised learning of graph neural networks: A unified review, 2021

    Y aochen Xie, Zhao Xu, Jingtun Zhang, Zhengyang Wang, and Shuiwang Ji. Self-supervised learning of graph neural networks: A unified review, 2021

    2021

  39. [47]

    Y uning Y ou, Tianlong Chen, Y ang Shen, and Zhangyang Wang, ‘Graph contrastive learning automated’,CoRR, abs/2106.07594, (2021)

    2021 arXiv

  40. [48]

    Y uning Y ou, Tianlong Chen, Y ongduo Sui, Ting Chen, Zhangyang Wang, and Y ang Shen, ‘Graph contrastive learning with augmentations’,CoRR, (2020)

    2020

  41. [49]

    Jure Zbontar, Li Jing, Ishan Misra, Y ann LeCun, and Stéphane Deny, ‘Barlow twins: Self-supervised learning via redundancy reduction’, (2021)

    2021

  42. [50]

    Graphsaint: Graph sampling based inductive learning method, 2020

    Hanqing Zeng, Hongkuan Zhou, Ajitesh Srivastava, Rajgopal Kannan, and Viktor Prasanna. Graphsaint: Graph sampling based inductive learning method, 2020

    2020

  43. [51]

    Y anqiao Zhu, Yichen Xu, Hejie Cui, Carl Y ang, Qiang Liu, and Shu Wu, ‘Structure-enhanced heterogeneous graph contrastive learning’, in Proceedings of the 2022 SIAM International Conference on Data Mining (SDM), pp. 82–90. SIAM, (2022)

    2022

  44. [52]

    Deep graph contrastive representation learning, 2020

    Y anqiao Zhu, Yichen Xu, Feng Y u, Qiang Liu, Shu Wu, and Liang Wang. Deep graph contrastive representation learning, 2020

    2020

  45. [53]

    2069–2080, New Y ork, NY , USA, (2021)

    Y anqiao Zhu, Yichen Xu, Feng Y u, Qiang Liu, Shu Wu, and Liang Wang, ‘Graph contrastive learning with adaptive augmentation’, inProceedings of WWW’21, WWW ’21, p. 2069–2080, New Y ork, NY , USA, (2021). ACM

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.