Pith. sign in

REVIEW 5 major objections 6 minor 1 cited by

Cluster-guided Contrastive Class-imbalanced Graph Classification

T0 review · 5 major / 6 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 The paper claims that C3GNN—adaptive clustering of majority classes into balanced subclasses, subclass Mixup, and hierarchical supervised contrastive learning—outperforms re-sampling, re-weighting, and long-tailed graph baselines on all…

desk verdict A plausible, clearly-written method for class-imbalanced graph classification that deserves a careful review, but the empirical case is undercut by missing error bars and per-dataset tuning. read the letter →

arxiv 2412.12984 v2 pith:F2YVB7UX submitted 2024-12-17 cs.LG cs.AIcs.IRcs.SI

Wei Ju , Zhengyang Mao , Siyu Yi , Yifang Qin , Yiyang Gu , Zhiping Xiao , Jianhao Shen , Ziyue Qiao
show 1 more author
Ming Zhang
This is my paper · ORCID
classification cs.LGcs.AIcs.IRcs.SI
keywords class-imbalancedgraphclassificationneuralnetworkssupervisedcontrastivelearningadaptiveclusteringMixuplong-tailedsubclassbalancinggraph-levelrepresentation
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

Class-imbalanced graph classification is hard because GNNs trained on skewed class distributions tilt toward the largest classes. The paper proposes C3GNN, which clusters each majority class into multiple subclasses whose sizes mimic the minority class, so that no single class dominates the learning budget. Inside each subclass, Mixup creates interpolated training graphs, and two supervised contrastive losses organize the representation space hierarchically: one pulls graphs in the same subclass together, the other pulls same-class but different-subclass graphs together. The paper reports that this combination outperforms all compared re-sampling, re-weighting, contrastive, and long-tailed graph baselines on all six benchmark datasets.

What carries the argument

The load-bearing mechanism is the adaptive subclass assignment produced by k-means on current GNN graph embeddings, with subclass size capped by $M = \max(n_K, \delta)$. This converts a skewed class distribution into a balanced set of pseudo-classes, which is what lets the two supervised contrastive losses (Eqs. 6 and 7) learn without letting majority classes dominate. Subclass Mixup (Eq. 5) guards against representation collapse in sparse subclasses, and cluster centers are refreshed every $T$ epochs so that the partition co-evolves with the encoder.

What would settle it

Run C3GNN with the same cluster sizes but with each majority class randomly partitioned into subclasses, and compare accuracy on all six datasets: if random partitions achieve nearly the same results as the reported k-means assignments, the claim that discovered semantic substructure drives the gains is refuted. A complementary check is to measure cluster stability across the $T$-epoch updates with the adjusted Rand index; low stability without a corresponding accuracy drop would also weaken the mechanism.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that majority classes in imbalanced graph data should be treated as collections of semantic subclasses rather than as monolithic classes. C3GNN encodes graphs with a GNN, applies k-means to the embeddings of each majority class to form subclasses capped at size $M = \max(n_K, \delta)$, synthesizes additional samples inside each subclass by Mixup interpolation, and optimizes the joint loss $\mathcal{L} = \sum_i (\mathcal{L}^{\mathrm{intra}}_i + \beta \mathcal{L}^{\mathrm{inter}}_i)$. The intra-subclass loss (Eq. 6) makes augmented views of graphs with the same subclass label attract each other, while the inter-subclass loss (Eq. 7) attracts graphs from different subclasses of the same class and repels graphs from other subclasses. In the reported experiments, this hierarchical structure yields the best top-1 accuracy on Synthie, ENZYMES, MNIST, Letter-high, Letter-low, and COIL-DEL across all tested imbalance factors.

Load-bearing premise

The method assumes that k-means on the current GNN embeddings of each majority class separates that class into semantically coherent subclasses, so that balancing those subclasses against the minority class is meaningful rather than arbitrary; if the clusters are unstable or do not match real substructure, the two contrastive losses would reinforce accidental partitions.

Editorial extensions

If this is right

  • On each of the six benchmark datasets, C3GNN reports the highest average top-1 accuracy among the compared methods, with the gap over baselines generally widening as the imbalance factor increases.
  • The ablation study reports that removing hierarchical subclass contrastive learning, adaptive clustering, or subclass Mixup each lowers accuracy on all six datasets, so the three modules are presented as jointly necessary for the result.
  • Feature-distance analysis on MNIST with imbalance factor 50 shows smaller intra-subclass distances than intra-class distances and more uniform subclass volumes across many, medium, and few-shot regions, indicating a subclass-balanced representation space.
  • The sensitivity analysis shows an optimal cluster size that grows when the minority class is larger, implying the method's preferred granularity is tied to minority-class size.

Reading between the lines

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

  • A testable extension outside the paper is to replace k-means with a supervised or structural sublabel discovery method and check whether accuracy tracks cluster semantic purity; if it does not, the reported gains may come mainly from rebalancing rather than from recovered substructure.
  • The approach is stated for graph classification, but the same cluster-then-hierarchical-contrast recipe could transfer to other long-tailed structured inputs with latent subcategories, such as point clouds or code snippets, since nothing in Eqs. (5)–(7) is graph-specific beyond the encoder.
  • Because C3GNN's contrastive objectives shape representations but do not alter the final classifier, stacking it with logit adjustment or class-balanced losses at the classification head is a natural combination the paper does not explore.
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

5 major / 6 minor

Summary. The paper proposes C3GNN, a method for class-imbalanced graph classification that combines adaptive k-means clustering of majority classes into balanced subclasses, subclass-level Mixup interpolation, and hierarchical supervised contrastive losses (intra-subclass and inter-subclass). The central claim is that C3GNN outperforms all considered baselines on all six benchmark datasets across various imbalance factors (IFs). The method is evaluated on synthetic and real-world graph classification datasets with GraphSAGE as the encoder, and ablations, hyperparameter sensitivity, feature-distance analyses, and a qualitative visualization of clusters are provided.

Significance. If the empirical claims are substantiated, the paper would make a useful contribution to the underexplored area of class-imbalanced graph classification. The central idea of decomposing majority classes into semantically coherent subclasses to balance class sizes and then applying hierarchical contrastive learning is simple and well motivated, and the paper combines several known components (clustering, Mixup, SupCon) in a novel configuration for the graph-classification setting. The experimental coverage includes six datasets and a range of baseline families (re-sampling, re-weighting, contrastive, and augmentation-based methods), which is a strength. However, the headline claim of universal superiority currently rests on point estimates without variance or significance testing, and the mechanism attribution is supported partly by a circular validation (the feature distances in Figure 3 are the direct objective of the losses in Eqs. 6-8). These gaps are load-bearing for the central claim and must be addressed before the results can be accepted at face value.

major comments (5)
  1. [Table 1 / Implementation details] Table 1 reports only average top-1 accuracy over 10 runs, with no standard deviations, per-seed results, or significance tests. Several margins over the best baseline are small (e.g., 0.68 pp on Letter-high IF=25 and 1.25 pp on Synthie IF=15 relative to RAHNet). Without uncertainty quantification, the claim that C3GNN "outperforms all baselines on all six datasets" is not statistically supported. Please report mean +/- std over the 10 runs and conduct paired tests (e.g., paired t-test or Wilcoxon signed-rank test) against the strongest baseline for each configuration.
  2. [Implementation details / Hyper-parameter Sensitivity (Figure 2)] The cluster size control parameter δ (Eq. 4) is "fine-tuned ... for each dataset individually," and Figure 2 shows accuracy variations of up to roughly 5 percentage points across cluster sizes (e.g., Synthie IF=15). If δ is selected using the test set or the same validation set used for early stopping, the reported gains may partly reflect per-dataset hyperparameter selection rather than the method's intrinsic advantage. Please clarify whether δ was chosen on a separate validation split, report the exact δ values used for each dataset and IF, and consider reporting results averaged over a range of δ or using a principled criterion (e.g., validation accuracy) to pick δ.
  3. [Datasets / Eq. (1)] The construction of the Zipf-imbalanced training sets is not fully specified. The text says the training sets were "transformed into class-imbalanced datasets with varying imbalance factors" and Eq. (1) gives a general integral condition, but the exact sampling procedure is missing: how many classes are included, how the class frequencies are assigned under Zipf's law, whether the 6:2:2 split is performed before or after the imbalance transformation, and whether the test/validation sets are truly balanced with the same original class labels. Without this specification the reported IF values and the comparability of results across methods cannot be fully assessed.
  4. [Analysis of Feature Distribution / Eqs. (6)-(8)] The validation of the clustering mechanism is circular. Figure 3 shows that intra-subclass distances are smaller than intra-class distances and that subclass distances are more uniform across many/medium/few regions; however, these properties are exactly what the losses in Eqs. (6)-(8) are designed to enforce, so the figure does not provide independent evidence that the clusters correspond to semantically meaningful substructures. Please add a non-circular evaluation, for example: (i) cluster stability across re-initializations or across repeated runs with different random seeds, (ii) comparison against random partitions of the same class into the same number of subclasses, or (iii) quantitative alignment with known semantic groupings if any exist in the datasets.
  5. [Visualization of Clustering Results / Algorithm 1] Figure 4 is a qualitative, hand-picked illustration, and the clustering is recomputed every T=10 epochs from a moving encoder, so the encoder and cluster assignments co-adapt. On its own this does not establish that the clusters capture stable semantic substructure; the gains could in principle come from arbitrary partitions combined with additional contrastive regularization. Please report a quantitative measure of cluster stability over training (e.g., adjusted Rand index between consecutive clusterings or agreement across different initializations) and, where possible, compare against a variant that uses fixed clusters computed once before training.
minor comments (6)
  1. [Introduction] The sentence "which aims to classify the class labels of graphs and has emerged as one of the most interesting also popular topics" contains a typo: "also popular" should be "and popular".
  2. [Equation (7)] The notation P(i)/Q(i) and A(i)/Q(i) is ambiguous; it should be written as P(i) \setminus Q(i) and A(i) \setminus Q(i) to denote set difference.
  3. [Figure 2] The x-axis is labeled "Cluster size," but Eq. (4) defines M as the threshold that indirectly controls cluster size via δ. Please clarify in the caption and text whether the x-axis is δ or the resulting number of subclasses per majority class, and explain how the displayed range corresponds to the tuning procedure.
  4. [Algorithm 1] The line "Sample one augmentation from You et al. (2020)" is underspecified; please state which of the four augmentations (node dropping, edge perturbation, attribute masking, subgraph) are used, and whether the augmentation is sampled per graph or per batch.
  5. [Experiments] The abstract and introduction mention NCI as a motivating example, but NCI is not among the six datasets in Table 1. Adding it or removing the motivational reference would improve consistency.
  6. [Computational Complexity Analysis] The complexity statement says the clustering complexity is O(I K N D) and "simplified to O(N D)" because I and K are constants; since K is the number of clusters, please clarify whether K is treated as bounded by the number of classes times a small constant, and state any assumptions about the number of majority classes.

Circularity Check

1 steps flagged · score 2.0 of 10

One circular feature-distribution validation; the benchmark accuracy claim itself is external and not circular.

  1. self definitional [Experiment > Analysis of Feature Distribution, discussion of Figure 3]
    "From the results, we can draw several conclusions: (i) The average intra-subclass distance is lower than the average intra-class distance, which implies that C 3GNN encourages samples from the same subclass to have similar representations. This observation suggests that C 3GNN successfully captures and emphasizes the finer distinctions present within each class."

    Eq. (6) defines the intra-subclass supervised contrastive loss to pull same-subclass representations together, and Eq. (7) pushes different subclasses within the same class apart. The reported 'intra-subclass distance < intra-class distance' is therefore a direct consequence of optimizing these losses on the same learned embeddings; it verifies that the objective was optimized rather than providing independent evidence that k-means discovered semantic graph substructure. No external cluster-purity or stability measure is supplied. This circularity is confined to the interpretability/validation figure and does not contaminate Table 1, where accuracy is measured against external benchmarks.

full rationale

The central claim of C3GNN is improved class-imbalanced graph classification accuracy, evaluated on six external benchmark datasets against baselines. No fitted parameter is renamed as a prediction, and the contrastive losses are not constructed to force the reported test accuracies. The method combines external components (Mixup, SupCon, graph augmentations, k-means) rather than importing a load-bearing result from the authors' prior work. One genuinely circular element exists in the feature-distribution analysis: Eqs. (6)-(8) explicitly optimize the same intra-subclass/inter-subclass distances that Figure 3 then reports as evidence of semantic substructure, making that validation self-confirming. This does not affect the external accuracy comparison, so the overall circularity score is low; per-dataset tuning of delta and missing error bars are correctness risks, not circularity.

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

The method relies on common ML components and a few hand-chosen hyperparameters. No new physical or mathematical entities are invented. The main unstated assumptions are about the semantic validity of k-means subclasses and the realism of the synthetic Zipf-imbalance evaluation.

free parameters (4)
  • delta (cluster size control) = tuned per dataset, values not reported
    Eq. (4) uses M = max(n_K, delta) to set the upper limit on subclass size; the paper fine-tunes delta individually per dataset and shows accuracy varies with cluster size in Figure 2.
  • temperature tau = 0.2
    Temperature in the supervised contrastive losses (Eqs. 6-7); chosen by hand, not justified.
  • update interval T = 10
    Cluster centers are recomputed every T epochs (Algorithm 1); fixed value, reported without sensitivity analysis.
  • beta = 1
    Balance weight between intra- and inter-subclass losses in Eq. (8); set to 1 by default, no tuning shown.
assumptions (4)
  • domain assumption K-means clustering on GNN embeddings of a majority class yields semantically coherent subclasses whose sizes can be balanced against the minority class.
    Used in the adaptive clustering module; if clusters are arbitrary, the hierarchical contrastive losses enforce meaningless partitions. Only qualitative evidence (Figure 4) is given.
  • domain assumption The benchmark construction (Zipf's law training distribution, balanced validation/test) captures realistic class-imbalanced graph classification.
    The evaluation section states training sets follow Zipf's law exactly, but the exact sampling procedure is not described.
  • domain assumption Standard graph augmentations (node dropping, edge perturbation, attribute masking, subgraph) preserve the graph's class and subclass label in contrastive learning.
    The augmentation distribution T(·|G) is taken from You et al. 2020 without empirical verification of label preservation on these datasets.
  • ad hoc to paper The threshold M = max(n_K, delta) is a reasonable ad hoc rule for balancing subclass sizes.
    No principled derivation or theoretical justification beyond balancing sample counts; appears as Eq. (4).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Cluster-guided Contrastive Class-imbalanced Graph Classification." pith.science (2026). https://pith.science/paper/F2YVB7UX

@misc{pith2026241212984,
  author       = {Pith},
  title        = {Pith review of: Cluster-guided Contrastive Class-imbalanced Graph Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/F2YVB7UX}},
  note         = {Machine review of arXiv:2412.12984}
}
abstract

This paper studies the problem of class-imbalanced graph classification, which aims at effectively classifying the graph categories in scenarios with imbalanced class distributions. While graph neural networks (GNNs) have achieved remarkable success, their modeling ability on imbalanced graph-structured data remains suboptimal, which typically leads to predictions biased towards the majority classes. On the other hand, existing class-imbalanced learning methods in vision may overlook the rich graph semantic substructures of the majority classes and excessively emphasize learning from the minority classes. To address these challenges, we propose a simple yet powerful approach called C$^3$GNN that integrates the idea of clustering into contrastive learning to enhance class-imbalanced graph classification. Technically, C$^3$GNN clusters graphs from each majority class into multiple subclasses, with sizes comparable to the minority class, mitigating class imbalance. It also employs the Mixup technique to generate synthetic samples, enriching the semantic diversity of each subclass. Furthermore, supervised contrastive learning is used to hierarchically learn effective graph representations, enabling the model to thoroughly explore semantic substructures in majority classes while avoiding excessive focus on minority classes. Extensive experiments on real-world graph benchmark datasets verify the superior performance of our proposed method against competitive baselines.

Figures

Figures reproduced from arXiv: 2412.12984 by the authors.

Figure 1
Figure 1. In the subsequent sections, we provide a detailed [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 1
Figure 1. Illustration of the proposed framework C [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. Feature distance and distribution curve of classes [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Visualization of clustering results. We visualize [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. CEGA: A Cost-Effective Approach for Graph-Based Model Extraction and Acquisition

    cs.LG 2025-06 conditional novelty 5.0 of 10

    CEGA iteratively selects nodes by balancing PageRank centrality, prediction entropy, and embedding diversity to replicate a target GNN's behavior on a limited query budget.

Reference graph

Works this paper leans on

56 extracted references · 37 canonical work pages · cited by 1 Pith paper

  1. [1]

    Cao, K.; Wei, C.; Gaidon, A.; Arechiga, N.; and Ma, T. 2019. Learning imbalanced datasets with label-distribution-aware margin loss. Advances in neural information processing systems, 32

    2019

  2. [2]

    Chawla, N. V. 2003. C4. 5 and imbalanced data sets: investigating the effect of sampling method, probabilistic estimate, and decision tree structure. In Proceedings of the ICML, volume 3, 66. CIBC Toronto, ON, Canada

    work page 2003

  3. [3]

    V.; Bowyer, K

    Chawla, N. V.; Bowyer, K. W.; Hall, L. O.; and Kegelmeyer, W. P. 2002. SMOTE: synthetic minority over-sampling technique. Journal of artificial intelligence research, 16: 321--357

    2002

  4. [4]

    Chen, J.; and Kou, G. 2023. Attribute and Structure Preserving Graph Contrastive Learning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 7024--7032

    work page 2023

  5. [5]

    Chen, T.; Kornblith, S.; Norouzi, M.; and Hinton, G. 2020. A simple framework for contrastive learning of visual representations. In International Conference on Machine Learning, 1597--1607. PMLR

    work page 2020

  6. [6]

    Cui, Y.; Jia, M.; Lin, T.-Y.; Song, Y.; and Belongie, S. 2019. Class-balanced loss based on effective number of samples. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 9268--9277

    2019

  7. [7]

    P.; Joshi, C

    Dwivedi, V. P.; Joshi, C. K.; Laurent, T.; Bengio, Y.; and Bresson, X. 2020. Benchmarking graph neural networks. arXiv preprint arXiv:2003.00982

    arXiv 2020

  8. [8]

    S.; Riley, P

    Gilmer, J.; Schoenholz, S. S.; Riley, P. F.; Vinyals, O.; and Dahl, G. E. 2017. Neural message passing for quantum chemistry. In International Conference on Machine Learning, 1263--1272. PMLR

    work page 2017

Show all 56 references
  1. [9]

    Gong, X.; Yang, C.; and Shi, C. 2023. Ma-gcl: Model augmentation tricks for graph contrastive learning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 4284--4292

    2023

  2. [10]

    Gu, Y.; Chen, Z.; Qin, Y.; Mao, Z.; Xiao, Z.; Ju, W.; Chen, C.; Hua, X.-S.; Wang, Y.; Luo, X.; et al. 2024. DEER: Distribution Divergence-based Graph Contrast for Partial Label Learning on Graphs. IEEE Transactions on Multimedia

    2024

  3. [11]

    Guo, H.; and Wang, S. 2021. Long-tailed multi-label visual recognition by collaborative training on uniform and re-balanced samplings. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 15089--15098

    2021

  4. [12]

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

    2017

  5. [13]

    Han, H.; Wang, W.-Y.; and Mao, B.-H. 2005. Borderline-SMOTE: a new over-sampling method in imbalanced data sets learning. In International conference on intelligent computing, 878--887. Springer

    2005

  6. [14]

    He, K.; Fan, H.; Wu, Y.; Xie, S.; and Girshick, R. 2020. Momentum contrast for unsupervised visual representation learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 9729--9738

    2020

  7. [15]

    Hou, C.; Zhang, J.; Wang, H.; and Zhou, T. 2023. Subclass-balancing contrastive learning for long-tailed recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 5395--5407

    2023

  8. [16]

    Ju, W.; Fang, Z.; Gu, Y.; Liu, Z.; Long, Q.; Qiao, Z.; Qin, Y.; Shen, J.; Sun, F.; Xiao, Z.; et al. 2024 a . A comprehensive survey on deep graph representation learning. Neural Networks, 106207

    2024

  9. [17]

    Ju, W.; Luo, X.; Qu, M.; Wang, Y.; Chen, C.; Deng, M.; Hua, X.-S.; and Zhang, M. 2022. TGNN: A joint semi-supervised framework for graph-level classification. In Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence, IJCAI, 2122--2128

    2022

  10. [18]

    Ju, W.; Mao, Z.; Yi, S.; Qin, Y.; Gu, Y.; Xiao, Z.; Wang, Y.; Luo, X.; and Zhang, M. 2024 b . Hypergraph-enhanced Dual Semi-supervised Graph Classification. In International Conference on Machine Learning. PMLR

    2024

  11. [19]

    Ju, W.; Wang, Y.; Qin, Y.; Mao, Z.; Xiao, Z.; Luo, J.; Yang, J.; Gu, Y.; Wang, D.; Long, Q.; et al. 2024 c . Towards Graph Contrastive Learning: A Survey and Beyond. arXiv preprint arXiv:2405.11868

    2024 arXiv

  12. [20]

    Ju, W.; Yi, S.; Wang, Y.; Xiao, Z.; Mao, Z.; Li, H.; Gu, Y.; Qin, Y.; Yin, N.; Wang, S.; et al. 2024 d . A survey of graph neural networks in real world: Imbalance, noise, privacy and ood challenges. arXiv preprint arXiv:2403.04468

    2024

  13. [21]

    Kashima, H.; Tsuda, K.; and Inokuchi, A. 2003. Marginalized kernels between labeled graphs. In Proceedings of International Conference on Machine Learning, 321--328

    2003

  14. [22]

    Khosla, P.; Teterwak, P.; Wang, C.; Sarna, A.; Tian, Y.; Isola, P.; Maschinot, A.; Liu, C.; and Krishnan, D. 2020. Supervised contrastive learning. Advances in Neural Information Processing Systems, 33: 18661--18673

    2020

  15. [23]

    N.; and Welling, M

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

    2016 arXiv

  16. [24]

    Liu, Z.; Nguyen, T.-K.; and Fang, Y. 2021. Tail-gnn: Tail-node graph neural networks. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, 1109--1119

    2021

  17. [25]

    Luo, J.; Gu, Y.; Luo, X.; Ju, W.; Xiao, Z.; Zhao, Y.; Yuan, J.; and Zhang, M. 2024 a . GALA: Graph Diffusion-based Alignment with Jigsaw for Source-free Domain Adaptation. IEEE Transactions on Pattern Analysis & Machine Intelligence, (01): 1--14

    2024

  18. [26]

    Luo, J.; Luo, X.; Chen, X.; Xiao, Z.; Ju, W.; and Zhang, M. 2024 b . SemiEvol: Semi-supervised Fine-tuning for LLM Adaptation. arXiv preprint arXiv:2410.14745

    2024 arXiv

  19. [27]

    Luo, X.; Ju, W.; Gu, Y.; Qin, Y.; Yi, S.; Wu, D.; Liu, L.; and Zhang, M. 2023 a . Toward effective semi-supervised node classification with hybrid curriculum pseudo-labeling. ACM Transactions on Multimedia Computing, Communications and Applications, 20(3): 1--19

    2023

  20. [28]

    Luo, X.; Zhao, Y.; Qin, Y.; Ju, W.; and Zhang, M. 2023 b . Towards semi-supervised universal graph classification. IEEE Transactions on Knowledge and Data Engineering, 36(1): 416--428

    2023

  21. [29]

    Mao, Z.; Ju, W.; Qin, Y.; Luo, X.; and Zhang, M. 2023. Rahnet: Retrieval augmented hybrid network for long-tailed graph classification. In Proceedings of the 31st ACM International Conference on Multimedia, 3817--3826

    2023

  22. [30]

    Mao, Z.; Ju, W.; Yi, S.; Wang, Y.; Xiao, Z.; Long, Q.; Yin, N.; Liu, X.; and Zhang, M. 2024. Learning Knowledge-diverse Experts for Long-tailed Graph Classification. ACM Transactions on Knowledge Discovery from Data

    2024

  23. [31]

    K.; Jayasumana, S.; Rawat, A

    Menon, A. K.; Jayasumana, S.; Rawat, A. S.; Jain, H.; Veit, A.; and Kumar, S. 2020. Long-tail learning via logit adjustment. arXiv preprint arXiv:2007.07314

    2020 arXiv

  24. [32]

    M.; Kersting, K.; and Mutzel, P

    Morris, C.; Kriege, N. M.; Kersting, K.; and Mutzel, P. 2016. Faster kernels for graphs with continuous attributes via hashing. In 2016 IEEE 16th International Conference on Data Mining (ICDM), 1095--1100. IEEE

    2016

  25. [33]

    Newman, M. E. 2005. Power laws, Pareto distributions and Zipf's law. Contemporary Physics, 46(5): 323--351

    2005

  26. [34]

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

    2018 arXiv

  27. [35]

    Park, J.; Song, J.; and Yang, E. 2021. GraphENS: Neighbor-Aware Ego Network Synthesis for Class-Imbalanced Node Classification. In International Conference on Learning Representations

    2021

  28. [36]

    Riesen, K.; and Bunke, H. 2008. IAM graph database repository for graph based pattern recognition and machine learning. In Joint IAPR International Workshops on Statistical Techniques in Pattern Recognition (SPR) and Structural and Syntactic Pattern Recognition (SSPR), 287--29...

    2008

  29. [37]

    Schomburg, I.; Chang, A.; Ebeling, C.; Gremse, M.; Heldt, C.; Huhn, G.; and Schomburg, D. 2004. BRENDA, the enzyme database: updates and major new developments. Nucleic acids research, 32(suppl\_1): D431--D433

    2004

  30. [38]

    J.; Mehlhorn, K.; and Borgwardt, K

    Shervashidze, N.; Schweitzer, P.; Van Leeuwen, E. J.; Mehlhorn, K.; and Borgwardt, K. M. 2011. Weisfeiler-lehman graph kernels. Journal of Machine Learning Research, 12(9): 2539--2561

    2011

  31. [39]

    Shervashidze, N.; Vishwanathan, S.; Petri, T.; Mehlhorn, K.; and Borgwardt, K. 2009. Efficient graphlet kernels for large graph comparison. In Proceedings of International Conference on Artificial Intelligence and Statistics, 488--495

    2009

  32. [40]

    Shi, M.; Tang, Y.; Zhu, X.; Wilson, D.; and Liu, J. 2020. Multi-class imbalanced graph convolutional network learning. In Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence (IJCAI-20)

    2020

  33. [41]

    Song, J.; Park, J.; and Yang, E. 2022. TAM: Topology-Aware Margin Loss for Class-Imbalanced Node Classification. In International Conference on Machine Learning, 20369--20383. PMLR

    2022

  34. [42]

    A.; and Karypis, G

    Wale, N.; Watson, I. A.; and Karypis, G. 2008. Comparison of descriptor spaces for chemical compound retrieval and classification. Knowledge and Information Systems, 14(3): 347--375

    2008

  35. [43]

    Wang, X.; Lian, L.; Miao, Z.; Liu, Z.; and Yu, S. X. 2020. Long-tailed recognition by routing diverse distribution-aware experts. arXiv preprint arXiv:2010.01809

    2020 arXiv

  36. [44]

    Wang, Y.; Zhao, Y.; Shah, N.; and Derr, T. 2022. Imbalanced graph classification via graph-of-graph neural networks. In Proceedings of the 31st ACM International Conference on Information & Knowledge Management, 2067--2076

    2022

  37. [45]

    Xiang, L.; Ding, G.; and Han, J. 2020. Learning from multiple experts: Self-paced knowledge distillation for long-tailed classification. In European Conference on Computer Vision, 247--263. Springer

    2020

  38. [46]

    Yi, S.; Ju, W.; Qin, Y.; Luo, X.; Liu, L.; Zhou, Y.; and Zhang, M. 2023 a . Redundancy-free self-supervised relational learning for graph clustering. IEEE Transactions on Neural Networks and Learning Systems

    2023

  39. [47]

    Yi, S.-Y.; Mao, Z.; Ju, W.; Zhou, Y.-D.; Liu, L.; Luo, X.; and Zhang, M. 2023 b . Towards long-tailed recognition for graph classification via collaborative experts. IEEE Transactions on Big Data

    2023

  40. [48]

    You, Y.; Chen, T.; Sui, Y.; Chen, T.; Wang, Z.; and Shen, Y. 2020. Graph contrastive learning with augmentations. Advances in Neural Information Processing Systems, 33: 5812--5823

    2020

  41. [49]

    N.; and Xia, F

    Yu, S.; Huang, H.; Dao, M. N.; and Xia, F. 2022. Graph augmentation learning. In Companion Proceedings of the Web Conference 2022, 1063--1072

    2022

  42. [50]

    Yun, S.; Kim, K.; Yoon, K.; and Park, C. 2022. LTE4G: Long-Tail Experts for Graph Neural Networks. In Proceedings of the 31st ACM International Conference on Information & Knowledge Management, 2434--2443

    2022

  43. [51]

    Zeng, L.; Li, L.; Gao, Z.; Zhao, P.; and Li, J. 2023. Imgcl: Revisiting graph contrastive learning on imbalanced node classification. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 11138--11146

    2023

  44. [52]

    N.; and Lopez-Paz, D

    Zhang, H.; Cisse, M.; Dauphin, Y. N.; and Lopez-Paz, D. 2017. mixup: Beyond empirical risk minimization. arXiv preprint arXiv:1710.09412

    2017 arXiv

  45. [53]

    Zhou, B.; Cui, Q.; Wei, X.-S.; and Chen, Z.-M. 2020. Bbn: Bilateral-branch network with cumulative learning for long-tailed visual recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 9719--9728

    2020

  46. [54]

    Zhou, M.; and Gong, Z. 2023. GraphSR: A Data Augmentation Algorithm for Imbalanced Node Classification. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 4954--4962

    2023

  47. [55]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  48. [56]

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