Pith. sign in

REVIEW 4 major objections 5 minor 44 references

Self-Reinforced Graph Contrastive Learning

T0 review · 4 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 Graph contrastive learning can improve by letting the model's own encoder select which augmented views to treat as positives.

desk verdict A plausible plug-in for GCL positive-pair selection with a mixed results table, a sign error in the EM derivation, and a consistency claim the paper's own data do not support. read the letter →

arxiv 2505.13650 v1 pith:CNBQHWJ3 submitted 2025-05-19 cs.LG cs.AI

Chou-Ying Hsieh , Chun-Fu Jang , Cheng-En Hsieh , Qian-Hui Chen , Sy-Yen Kuo This is my paper · ORCID
classification cs.LGcs.AI
keywords graphneuralnetworkcontrastivelearningexpectation-maximizationalgorithmmanifoldhypothesisself-reinforcementpositivepairselectionclassification
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 quality of positive pairs, not just the augmentation recipe, is what limits graph contrastive learning, and that the model's own encoder can judge pair quality through latent-space distance. It introduces SRGCL, a plug-in module that generates many candidate augmented views, keeps only those closest to the anchor under a manifold-smoothness criterion, and gradually sharpens this choice as training proceeds. The loop is self-reinforcing: better positives make a better encoder, and a better encoder recognizes better positives. The reported experiments support this by showing gains over the base GraphCL and AutoGCL pipelines on most of eight graph-classification benchmarks, at the cost of extra compute and memory.

What carries the argument

The load-bearing components are the Unified Positive Pair Generator (UPPG) and the Manifold-inspired Positive Pair Selector (MiPPS). UPPG draws c candidate views for each anchor from a pool of rule-based and learnable augmentation strategies, and MiPPS scores each candidate by Euclidean distance in the encoder's latent space, retaining the k nearest under a softmax distribution whose temperature decays as training progresses. This selector operationalizes the manifold hypothesis, stated as the assumption that points close in the learned representation space are more likely to share a label, and it is the mechanism that converts the contrastive objective into a self-reinforcing cycle. The paper also recasts the alternating selection and contrastive update as an instance of stochastic expectation-maximization, with the E-step assigning candidates as presumed positives and the M-step updating the encoder on those assignments.

What would settle it

On a benchmark such as NCI1, record the fraction of top-k selected candidates that share the anchor's ground-truth label at each epoch. If SRGCL's accuracy improves while this same-class precision stays at or below chance, the self-reinforcing selection is not the operative mechanism; if precision starts above chance and rises with accuracy, the mechanism is confirmed.

Watch

Extended reading notes

Core claim

The central claim is that a graph contrastive learner can be improved by letting its own encoder choose which augmented views to treat as positives, and this choice improves the encoder in turn. The authors call this loop self-reinforced: SRGCL generates c candidate views through several augmentation strategies, encodes them with the shared GNN, and keeps the k nearest candidates in Euclidean distance as the positive set. To avoid committing to possibly wrong close neighbors early on, selection is probabilistic with a Boltzmann temperature that decays over training, and the whole procedure is framed as stochastic expectation-maximization with a latent same-class indicator per candidate. On eight standard graph-classification benchmarks the paper reports that GraphCL with SRGCL beats the base model on six datasets, and AutoGCL with probabilistic SRGCL beats its base model on six datasets.

Load-bearing premise

The argument assumes that, early in training, graphs that are close in the encoder's latent space are usually in the same class, so the top-k selector's choices are mostly true positives rather than systematic false ones.

Editorial extensions

If this is right

  • Adding SRGCL to an existing contrastive pipeline should improve downstream graph classification accuracy whenever the encoder's latent distances track meaningful graph similarity.
  • The probabilistic selection with decaying temperature carries part of the gain: AutoGCL with it beats its base model on six datasets, while AutoGCL without it beats its base model on only one.
  • The module needs no labels and does not change the contrastive loss, so it can be layered onto existing graph contrastive methods as a drop-in component.
  • The gains come with overhead: runtime rises with candidate count when views are generated on the host, and GPU-side generation raises memory use as the candidate set size grows.
  • The optimal temperature constant varies by dataset, so dataset-specific tuning is needed to realize the reported improvements reliably.

Reading between the lines

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

  • If the self-reinforcement claim is right, the fraction of selected candidates that share the anchor's class should rise as training progresses; measuring that fraction directly would turn the loop's premise into an observable quantity.
  • The same selector could be ported to node-level or link-level contrastive objectives, where the manifold-smoothness assumption may be weaker; those settings would stress-test whether latent distance is a reliable same-class proxy outside graph-level tasks.
  • The paper's EM derivation in Eq. (15) states the same-class probability as proportional to the raw distance, which is the wrong direction given its own smoothness assumption; reading it as inversely proportional would make the theory match the top-k selection, and a corrected derivation would remove an internal inconsistency.
  • Because gains vary by base model and by the temperature constant, the broad claim of consistent improvement likely depends on per-dataset tuning; a fixed-configuration evaluation across many datasets would clarify how much of the gain comes from the mechanism rather than from hyperparameter selection.
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

4 major / 5 minor

Summary. The paper introduces SRGCL, a plug-in module for graph contrastive learning that generates multiple candidate augmented views via a Unified Positive Pair Generator (UPPG) and then selects top-k positives using a Manifold-inspired Positive Pair Selector (MiPPS), with a probabilistic (temperature-decayed) variant SRGCL(p). Section III builds the framework on the manifold hypothesis: candidates closer to the anchor in the encoder's latent space are assumed more likely to be same-class positives. Section IV frames the procedure as stochastic expectation-maximization (SEM), alternating between hard assignment of positives and a contrastive M-step. Experiments on eight TUDataset benchmarks integrate SRGCL into GraphCL and AutoGCL, reporting accuracy for deterministic and probabilistic variants, plus ablations over distance functions, temperature constants, augmentation combinations, candidate set sizes, and runtime/memory overhead. The central claim in the abstract is that SRGCL 'consistently outperforms state-of-the-art GCL methods.'

Significance. If the claimed improvements were robust, a self-reinforcing positive-pair selector would be a useful plug-in contribution: it is model-agnostic, requires no labels, and is accompanied by released code and a reasonable set of ablations (distance functions, temperature schedule, candidate set size, overhead). The probabilistic optimizer with Boltzmann sampling and temperature decay is a sensible mechanism, and the framing as SEM is conceptually interesting. However, the current evidence does not support the abstract's blanket claim of consistent superiority: the deterministic AutoGCL_SR variant is worse than AutoGCL on seven of eight datasets, and GraphCL_SR(p) degrades on four of eight. The internal inconsistency in Eq. (15) and the lack of validation that the encoder's early latent geometry really selects same-class neighbors are load-bearing gaps. The core idea is defensible but requires substantial additional evidence and a more measured claim.

major comments (4)
  1. [Section IV-B, Eq. (15)] Equation (15) states p(z_ij = 1 | g_i, g'_ij, θ) ∝ D(f_θ(g_i), f_θ(g'_ij)), i.e., probability proportional to distance. This is the opposite of the manifold-smoothness assumption stated in Section III-B, where smaller distances are asserted to indicate a higher chance of same-class membership. The Boltzmann selector in Eq. (10) uses exp(-D/T), which is consistent with inverse distance, but the EM E-step text and Eq. (15) are not. The derivation is therefore internally inconsistent and must be corrected, or the surrounding text must be rewritten to state that larger similarity (smaller distance) corresponds to higher positive probability.
  2. [Table I and abstract claim] The abstract states that SRGCL 'consistently outperforms state-of-the-art GCL methods,' but Table I shows AutoGCL_SR (deterministic top-k) is worse than AutoGCL on seven of eight datasets (e.g., 81.62±0.37 vs 82.00±0.29 on NCI1; 68.20±3.18 vs 70.12±0.68 on COLLAB; 84.86±5.06 vs 88.58±1.49 on RDT-B; 52.64±9.37 vs 56.75±0.18 on RDT-M5K). GraphCL_SR(p) also degrades relative to GraphCL on four datasets (NCI1, PROTEINS, RDT-B, RDT-M5K). The claim of consistent improvement is contradicted by the authors' own table. Please either narrow the claim (e.g., 'often improves, with the probabilistic variant helping particularly for AutoGCL') or provide statistical tests (per-dataset paired comparisons across seeds) and an explanation of when the method helps or hurts.
  3. [Section III-B and self-reinforcement bootstrapping] The load-bearing premise is that the encoder's latent distances are informative enough early in training for top-k selection to pick same-class candidates. The paper does not provide any per-dataset validation of this premise. I request an analysis of the precision of the top-k selected positives (e.g., fraction of selected candidates that share the anchor's class, using validation labels or a labeled probe) as a function of training epoch, for representative datasets in each domain. Without such evidence, the gains of AutoGCL_SR(p) could be attributed to added augmentation diversity, stochastic regularization from the Boltzmann sampling, or the temperature schedule rather than to manifold-based selection. The paper's own AutoGCL_SR results suggest that deterministic selection can reinforce the wrong alignment on several datasets, which strengthens the need for this validation.
  4. [Section IV-B, convergence claim] The paper states, without proof, that 'Under mild regularity conditions, the sequence θ(t) converges to a stationary point of the SEM objective.' No such conditions are stated, and the E-step uses a hard top-k assignment that changes the effective optimization landscape as the encoder evolves, while the temperature T(t) is also time-varying. The convergence claim is not justified by the cited SEM literature, which typically assumes fixed posterior computations and standard regularity conditions. Either provide a formal statement with conditions or soften the claim to describe the procedure as an EM-style heuristic.
minor comments (5)
  1. [Abstract] The abstract contains a typo: 'an unified' should be 'a unified.'
  2. [Section IV-A heading] The heading 'General Expectation-Maximiation Algorithm' misspells 'Maximization.'
  3. [Figure 5 caption] The caption says 'Graph SR (p)' but the body text refers to 'GraphCL SR (p)'; please make the caption consistent with the method name.
  4. [Section III-A, Eq. (4)] The notation T_{Q_j} is inconsistent with the random variable κ_j introduced in Eq. (3); should be T_{κ_j}.
  5. [Section V-B] The sentence 'Subgraph augmentation is excluded due to its consistently large induced distance from the original graph' is presented without quantitative support; a brief experiment or a reference would help readers judge whether this exclusion biases the UPPG comparison.

Circularity Check

1 steps flagged · score 5.0 of 10

Selector and training objective share the same encoder-distance criterion; the self-reinforcement rationale is partly self-confirming, though external SVC benchmarks keep the headline accuracy claim from being fully circular.

  1. self definitional [Section III-B (Eq. 8) and Section IV-B (Eq. 18)]
    "the M-step reduces to standard contrastive learning over the top- k closest candidate views."

    The selected positive set S_i is defined in Eq. (8)/(16) as the k candidates with the smallest encoder-space distance D(e(g'), e(g)) under the current encoder. The M-step in Eq. (18) then updates the encoder by maximizing a contrastive loss over exactly that selected set, which pulls those same candidates closer to the anchor. Thus the encoder is directly trained to agree with its own earlier ranking, so the paper's claim that 'high-quality positives iteratively refine the encoder, and the strengthened encoder ... more accurately judges pair quality' is a self-confirming loop: 'quality' is operationalized as proximity in the very encoder that the loss then optimizes.

full rationale

The main empirical claim, that SRGCL as a plug-in module improves graph classification, is tested on external labels via a downstream SVC in Table I, so that claim is not forced by construction. The circularity is partial and located in the method's self-reinforcement rationale: 'high-quality' positive pairs are defined as the k nearest candidates in the current encoder's latent space, and the training objective then minimizes distance for exactly those candidates, making the claimed improvement in 'pair-quality discrimination' self-confirming. Eq. (15) also states p(z_ij=1) proportional to D(...), which is internally inconsistent with the inverse-distance selection actually used; this is a correctness flaw rather than a circularity. No load-bearing self-citations were found: the manifold-hypothesis references are external, and the comparison baselines come from prior work by others. Overall, the score reflects one substantial self-definitional loop in the theoretical framing, while the external benchmark evaluation prevents a higher circularity score.

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

The method rests on one central domain assumption (manifold smoothness), several standard ML assumptions, and several tuned hyperparameters (c, k, s, T0, distance metric). No new physical entities are introduced.

free parameters (5)
  • candidate set size c = 50
    Number of augmented views generated per anchor; set to 50 in all main experiments. Ablation (Fig. 6) shows accuracy and variance depend on c, especially for AutoGCL on RDT-B.
  • top-k positive count k = 2
    Number of closest candidates kept as positives; chosen without ablation in the paper.
  • temperature decay constant s = 0.4 for GraphCL_SR, 0.8 for AutoGCL_SR
    Controls how quickly the softmax selector becomes deterministic. Ablation (Fig. 4) shows optimal s varies by dataset, so the reported values are data-dependent.
  • initial temperature T0 = not specified
    T(t) = T0*exp(-s t) uses T0 but its value is never stated in Section III-C; likely default 1, making the schedule underdetermined.
  • distance metric D = Euclidean (L2)
    Chosen after comparing L2, cosine, KL, and Wasserstein; Fig. 3 shows Wasserstein performs poorly, so selecting L2 is a hand-picked design choice after observing results.
assumptions (5)
  • domain assumption Manifold hypothesis: graph-level data lie near a low-dimensional manifold in the learned embedding space.
    Invoked in Section III-B to justify using Euclidean distances in Z as a proxy for same-class membership; cited to prior work but not validated on the eight datasets.
  • domain assumption Manifold smoothness: points close in Z are more likely to have the same label.
    Section III-B and Eq. (8): the top-k selector relies on this to treat nearest latent neighbors as high-quality positives. This is the weakest structural premise.
  • standard math Contrastive losses such as InfoNCE induce clustered, class-discriminative latent spaces.
    Section III-B: assumed so that distance-based selection becomes more reliable as training proceeds; standard in contrastive learning literature.
  • ad hoc to paper Under mild regularity conditions, the SEM iterates converge to a stationary point.
    Section IV-B: asserted without stating the conditions or a proof; the paper supplies no convergence analysis for the proposed selection objective.
  • standard math The encoder is differentiable and locally linear, as used in the Jacobian argument of Eq. (9).
    Used to motivate label stability under small perturbations; a standard differentiability assumption, but not stated explicitly.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Self-Reinforced Graph Contrastive Learning." pith.science (2026). https://pith.science/paper/CNBQHWJ3

@misc{pith2026250513650,
  author       = {Pith},
  title        = {Pith review of: Self-Reinforced Graph Contrastive Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CNBQHWJ3}},
  note         = {Machine review of arXiv:2505.13650}
}
read the original abstract

Graphs serve as versatile data structures in numerous real-world domains-including social networks, molecular biology, and knowledge graphs-by capturing intricate relational information among entities. Among graph-based learning techniques, Graph Contrastive Learning (GCL) has gained significant attention for its ability to derive robust, self-supervised graph representations through the contrasting of positive and negative sample pairs. However, a critical challenge lies in ensuring high-quality positive pairs so that the intrinsic semantic and structural properties of the original graph are preserved rather than distorted. To address this issue, we propose SRGCL (Self-Reinforced Graph Contrastive Learning), a novel framework that leverages the model's own encoder to dynamically evaluate and select high-quality positive pairs. We designed a unified positive pair generator employing multiple augmentation strategies, and a selector guided by the manifold hypothesis to maintain the underlying geometry of the latent space. By adopting a probabilistic mechanism for selecting positive pairs, SRGCL iteratively refines its assessment of pair quality as the encoder's representational power improves. Extensive experiments on diverse graph-level classification tasks demonstrate that SRGCL, as a plug-in module, consistently outperforms state-of-the-art GCL methods, underscoring its adaptability and efficacy across various domains.

Figures

Figures reproduced from arXiv: 2505.13650 by the authors.

Figure 1
Figure 1. The core idea of SRGCL is that high-quality positive [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The proposed framework of SRGCL, which is the plug-in module for existing GCL frameworks. SRGCL comprises: [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Performance versus different distance functions across different datasets on Graph [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Performance versus temperature constant s on GraphCL SR (a) and AutoGCL SR (b). The light blue tunnel part is the standard deviation of SRGCL run in 5 times. The orange dashed line is the average accuracy of SRGCL without probabilistic optimization. Once s = 0, the SRG…
Figure 5
Figure 5. Figure 5: Performance versus augmentation combination across [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 7
Figure 7. Figure 7: Normalized runtime of GraphCL (a) and AutoGCL (b) [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 20 canonical work pages

  1. [1]

    Information network or social network? the structure of the twitter follow graph,

    S. A. Myers, A. Sharma, P. Gupta, and J. Lin, “Information network or social network? the structure of the twitter follow graph,” in Proceedings of the 23rd international conference on world wide web , 2014, pp. 493– 498

    work page 2014

  2. [2]

    Protein function prediction via graph ker- nels,

    K. M. Borgwardt, C. S. Ong, S. Sch ¨onauer, S. Vishwanathan, A. J. Smola, and H.-P. Kriegel, “Protein function prediction via graph ker- nels,” Bioinformatics, vol. 21, no. suppl 1, pp. i47–i56, 2005

    work page 2005

  3. [3]

    Knowledge graphs,

    A. Hogan, E. Blomqvist, M. Cochez, C. d’Amato, G. D. Melo, C. Gutierrez, S. Kirrane, J. E. L. Gayo, R. Navigli, S. Neumaier et al., “Knowledge graphs,” ACM Computing Surveys (Csur) , vol. 54, no. 4, pp. 1–37, 2021

    work page 2021

  4. [4]

    Graph neural networks for social recommendation,

    W. Fan, Y . Ma, Q. Li, Y . He, E. Zhao, J. Tang, and D. Yin, “Graph neural networks for social recommendation,” in The world wide web conference, 2019, pp. 417–426

    work page 2019

  5. [5]

    Graph con- trastive learning with augmentations,

    Y . You, T. Chen, Y . Sui, T. Chen, Z. Wang, and Y . Shen, “Graph con- trastive learning with augmentations,” Advances in neural information processing systems, vol. 33, pp. 5812–5823, 2020

    2020

  6. [6]

    Contrastive multi-view represen- tation learning on graphs,

    K. Hassani and A. H. Khasahmadi, “Contrastive multi-view represen- tation learning on graphs,” in International conference on machine learning. PMLR, 2020, pp. 4116–4126

    2020

  7. [7]

    Infograph: Unsupervised and semi-supervised graph-level representation learning via mutual information maximization,

    F.-Y . Sun, J. Hoffmann, V . Verma, and J. Tang, “Infograph: Unsupervised and semi-supervised graph-level representation learning via mutual information maximization,” arXiv preprint arXiv:1908.01000 , 2019

    arXiv 1908

  8. [8]

    Simgrace: A simple framework for graph contrastive learning without data augmentation,

    J. Xia, L. Wu, J. Chen, B. Hu, and S. Z. Li, “Simgrace: A simple framework for graph contrastive learning without data augmentation,” in Proceedings of the ACM Web Conference 2022 , 2022, pp. 1070–1079

    2022

Show all 44 references
  1. [9]

    Graph contrastive learning automated,

    Y . You, T. Chen, Y . Shen, and Z. Wang, “Graph contrastive learning automated,” in International Conference on Machine Learning. PMLR, 2021, pp. 12 121–12 132

    2021

  2. [10]

    Adversarial graph augmentation to improve graph contrastive learning,

    S. Suresh, P. Li, C. Hao, and J. Neville, “Adversarial graph augmentation to improve graph contrastive learning,” Advances in Neural Information Processing Systems, vol. 34, pp. 15 920–15 933, 2021

    2021

  3. [11]

    Autogcl: Automated graph contrastive learning via learnable view generators,

    Y . Yin, Q. Wang, S. Huang, H. Xiong, and X. Zhang, “Autogcl: Automated graph contrastive learning via learnable view generators,” in Proceedings of the AAAI conference on artificial intelligence , vol. 36, no. 8, 2022, pp. 8892–8900

    2022

  4. [12]

    Structure-activity relationship of mutagenic aromatic and heteroaromatic nitro compounds. correlation with molecular orbital energies and hydrophobicity,

    A. K. Debnath, R. L. Lopez de Compadre, G. Debnath, A. J. Shusterman, and C. Hansch, “Structure-activity relationship of mutagenic aromatic and heteroaromatic nitro compounds. correlation with molecular orbital energies and hydrophobicity,” Journal of medicinal chemistry , vol...

    1991

  5. [13]

    Momentum contrast for unsupervised visual representation learning,

    K. He, H. Fan, Y . Wu, S. Xie, and R. Girshick, “Momentum contrast for unsupervised visual representation learning,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 9729–9738

    2020

  6. [14]

    A simple framework for contrastive learning of visual representations,

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” in International conference on machine learning . PMLR, 2020, pp. 1597–1607

    2020

  7. [15]

    Representation learning with contrastive predictive coding,

    A. v. d. Oord, Y . Li, and O. Vinyals, “Representation learning with contrastive predictive coding,” arXiv preprint arXiv:1807.03748 , 2018

    2018 arXiv

  8. [16]

    A global geometric framework for nonlinear dimensionality reduction,

    J. B. Tenenbaum, V . d. Silva, and J. C. Langford, “A global geometric framework for nonlinear dimensionality reduction,” science, vol. 290, no. 5500, pp. 2319–2323, 2000

    2000

  9. [17]

    Representation learning: A review and new perspectives,

    Y . Bengio, A. Courville, and P. Vincent, “Representation learning: A review and new perspectives,” IEEE transactions on pattern analysis and machine intelligence , vol. 35, no. 8, pp. 1798–1828, 2013

    2013

  10. [18]

    Manifold structure in graph embeddings,

    P. Rubin-Delanchy, “Manifold structure in graph embeddings,” Advances in neural information processing systems , vol. 33, pp. 11 687–11 699, 2020

    2020

  11. [19]

    A manifold perspective on the statistical generalization of graph neural networks,

    Z. Wang, J. Cervino, and A. Ribeiro, “A manifold perspective on the statistical generalization of graph neural networks,” arXiv preprint arXiv:2406.05225, 2024

    2024 arXiv

  12. [20]

    An experimental study of the decay of temperature fluctuations in grid-generated turbulence,

    Z. Warhaft and J. Lumley, “An experimental study of the decay of temperature fluctuations in grid-generated turbulence,” Journal of Fluid Mechanics, vol. 88, no. 4, pp. 659–684, 1978

    1978

  13. [21]

    P. J. Van Laarhoven, E. H. Aarts, P. J. van Laarhoven, and E. H. Aarts, Simulated annealing. Springer, 1987

    1987

  14. [22]

    On upstream differencing and godunov-type schemes for hyperbolic conservation laws,

    A. Harten, P. D. Lax, and B. v. Leer, “On upstream differencing and godunov-type schemes for hyperbolic conservation laws,” SIAM review, vol. 25, no. 1, pp. 35–61, 1983

    1983

  15. [23]

    Statistical guarantees for the em algorithm: From population to sample-based analysis,

    S. Balakrishnan, M. J. Wainwright, and B. Yu, “Statistical guarantees for the em algorithm: From population to sample-based analysis,” 2017

    2017

  16. [24]

    Weakly- and semi-supervised learning of a deep convolutional network for semantic image segmentation,

    G. Papandreou, L.-C. Chen, K. P. Murphy, and A. L. Yuille, “Weakly- and semi-supervised learning of a deep convolutional network for semantic image segmentation,” in Proceedings of the IEEE international conference on computer vision , 2015, pp. 1742–1750

    2015

  17. [25]

    High-dimensional variance- reduced stochastic gradient expectation-maximization algorithm,

    R. Zhu, L. Wang, C. Zhai, and Q. Gu, “High-dimensional variance- reduced stochastic gradient expectation-maximization algorithm,” in International Conference on Machine Learning . PMLR, 2017, pp. 4180–4188

    2017

  18. [26]

    Tudataset: A collection of benchmark datasets for learning with graphs,

    C. Morris, N. M. Kriege, F. Bause, K. Kersting, P. Mutzel, and M. Neu- mann, “Tudataset: A collection of benchmark datasets for learning with graphs,” arXiv preprint arXiv:2007.08663 , 2020

    2007 arXiv

  19. [27]

    Pytorch: An imperative style, high-performance deep learning library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga et al. , “Pytorch: An imperative style, high-performance deep learning library,” Advances in neural information processing systems , vol. 32, 2019

    2019

  20. [28]

    Fast graph representation learning with pytorch geometric,

    M. Fey and J. E. Lenssen, “Fast graph representation learning with pytorch geometric,” arXiv preprint arXiv:1903.02428 , 2019

    1903 arXiv

  21. [29]

    Learning convolutional neural networks for graphs,

    M. Niepert, M. Ahmed, and K. Kutzkov, “Learning convolutional neural networks for graphs,” in International conference on machine learning . PMLR, 2016, pp. 2014–2023

    2016

  22. [30]

    Support vector machines for classifi- cation and regression,

    R. G. Brereton and G. R. Lloyd, “Support vector machines for classifi- cation and regression,” Analyst, vol. 135, no. 2, pp. 230–267, 2010

    2010

  23. [31]

    Efficient graphlet kernels for large graph comparison,

    N. Shervashidze, S. Vishwanathan, T. Petri, K. Mehlhorn, and K. Borg- wardt, “Efficient graphlet kernels for large graph comparison,” in Arti- ficial intelligence and statistics . PMLR, 2009, pp. 488–495

    2009

  24. [32]

    Weisfeiler-lehman graph kernels

    N. Shervashidze, P. Schweitzer, E. J. Van Leeuwen, K. Mehlhorn, and K. M. Borgwardt, “Weisfeiler-lehman graph kernels.” Journal of Machine Learning Research , vol. 12, no. 9, 2011

    2011

  25. [33]

    Deep graph kernels,

    P. Yanardag and S. Vishwanathan, “Deep graph kernels,” in Proceedings of the 21th ACM SIGKDD international conference on knowledge discovery and data mining , 2015, pp. 1365–1374

    2015

  26. [34]

    Semi-supervised classification with graph convolutional networks,

    T. N. Kipf and M. Welling, “Semi-supervised classification with graph convolutional networks,” arXiv preprint arXiv:1609.02907 , 2016

    2016 arXiv

  27. [35]

    Inductive representation learning on large graphs,

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

    2017

  28. [36]

    How powerful are graph neural networks?

    K. Xu, W. Hu, J. Leskovec, and S. Jegelka, “How powerful are graph neural networks?” arXiv preprint arXiv:1810.00826 , 2018

    2018 arXiv

  29. [37]

    Graph attention networks,

    P. Velickovic, G. Cucurull, A. Casanova, A. Romero, P. Lio, Y . Bengio et al., “Graph attention networks,” stat, vol. 1050, no. 20, pp. 10–48 550, 2017

    2017

  30. [38]

    node2vec: Scalable feature learning for networks,

    A. Grover and J. Leskovec, “node2vec: Scalable feature learning for networks,” in Proceedings of the 22nd ACM SIGKDD international conference on Knowledge discovery and data mining , 2016, pp. 855– 864

    2016

  31. [39]

    Sub2vec: Feature learning for subgraphs,

    B. Adhikari, Y . Zhang, N. Ramakrishnan, and B. A. Prakash, “Sub2vec: Feature learning for subgraphs,” in Advances in Knowledge Discov- ery and Data Mining: 22nd Pacific-Asia Conference, PAKDD 2018, Melbourne, VIC, Australia, June 3-6, 2018, Proceedings, Part II 22 . Springer,...

    2018

  32. [40]

    graph2vec: Learning distributed representations of graphs,

    A. Narayanan, M. Chandramohan, R. Venkatesan, L. Chen, Y . Liu, and S. Jaiswal, “graph2vec: Learning distributed representations of graphs,” arXiv preprint arXiv:1707.05005 , 2017

    2017 arXiv

  33. [41]

    A fair compari- son of graph neural networks for graph classification,

    F. Errica, M. Podda, D. Bacciu, and A. Micheli, “A fair compari- son of graph neural networks for graph classification,” arXiv preprint arXiv:1912.09893, 2019

    1912 arXiv

  34. [42]

    Autogcl github repository

    Yin, Yihang and Wang, Qingzhong and Huang, Siyu and Xiong, Haoyi and Zhang, Xiang. Autogcl github repository. [Online]. Available: https://github.com/Somedaywilldo/AutoGCL

  35. [43]

    I-divergence geometry of probability distributions and min- imization problems,

    I. Csisz ´ar, “I-divergence geometry of probability distributions and min- imization problems,” The annals of probability , pp. 146–158, 1975

    1975

  36. [44]

    Markov processes over denumerable products of spaces, describing large systems of automata,

    L. N. Vaserstein, “Markov processes over denumerable products of spaces, describing large systems of automata,” Problemy Peredachi Informatsii, vol. 5, no. 3, pp. 64–72, 1969

    1969

Pith tools

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