Pith. sign in

REVIEW 4 major objections 3 minor 48 references

SaGIF: Improving Individual Fairness in Graph Neural Networks via Similarity Encoding

T0 review · 4 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper claims that GNN individual unfairness is caused by feature–topology similarity inconsistency, and that a dedicated fused-similarity encoder can fix it without sacrificing utility.

desk verdict A reproducible method with a genuinely new idea, undermined by an evaluation oracle that doesn't match its training objective; fixable but the claims need tempering. read the letter →

arxiv 2506.18696 v1 pith:2TNFWFK3 submitted 2025-06-23 cs.LG

Yuchang Zhu , Jintang Li , Huizhe Zhang , Liang Chen , Zibin Zheng This is my paper
classification cs.LG
keywords individualfairnessgraphneuralnetworkssimilarityconsistencyencodingoraclematrixnodeclassificationtopologyfusionfeature
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 is trying to establish that individual unfairness in graph neural networks (GNNs) comes from a specific, measurable source: disagreement between the graph's feature-based notion of similarity and its topology-based notion of similarity. It introduces the concept of similarity consistency to quantify this disagreement, reports a correlation between low consistency and worse individual fairness, and treats that discrepancy as the cause to be fixed. Its fix is two fused oracle similarity metrics, topology fusion and feature fusion, that define 'similar individuals' using both feature and structural evidence. On top of these, SaGIF learns a separate similarity representation and concatenates it with the main GNN representation, which the paper shows improves ranking-based individual fairness metrics while retaining accuracy. A reader should care because if this is right, fairness in graphs becomes a similarity-measurement problem rather than a penalty-tuning problem.

What carries the argument

The mechanism that carries the argument is the similarity encoder: an independent graph-convolutional channel that takes as its layer-0 input the Laplacian eigenvector $p_i^0 \in \mathbb{R}^{d_{\mathrm{sim}}}$ of a kNN graph built from the fused oracle similarity matrix $S$, then passes those vectors through the same adjacency as the main encoder. At each layer the main encoder consumes concatenated $[h^{l-1}_i, p^{l-1}_i]$ inputs, so fused similarity information is always available to the classifier. The similarity channel is trained by $L_s(P,S) = \frac{1}{k|V_{\mathrm{train}}|} \|D(P,P^T)-S\|_F^2$, where $D$ is cosine similarity, which makes the learned similarities reproduce the fused oracle rather than relying on a hand-chosen Lipschitz penalty. This construction is what lets the paper claim fairness gains without changing the downstream task objective.

What would settle it

Compute NDCG@10 for SaGIF and baselines on the same six graphs using both the feature-only oracle and a fused topology-feature oracle. If SaGIF's advantage disappears when the ground truth is the fused oracle it was actually trained against, the paper's fairness claim is an artifact of mismatched measurement. Alternatively, artificially raise similarity consistency by rewiring edges among feature-similar nodes; if NDCG@10 does not rise, the claimed causal link fails.

Watch

Extended reading notes

Core claim

The central discovery, stated on the paper's terms, is that the predictions of a GNN become individually unfair because the nodes it treats as similar through message passing are not the nodes that are actually similar when feature and structural evidence are combined. The paper defines similarity consistency as the fraction of top-k nearest neighbors shared between feature-based and topology-based similarity lists, and its preliminary experiments show that low consistency tracks low NDCG@10 in node classification. It then proposes two ways of computing a fused oracle similarity matrix, one that blends the normalized adjacency with a feature kNN graph and one that embeds shortest-path distances into a synthetic feature matrix, and builds SaGIF around them. SaGIF preserves that fused similarity with a dedicated encoder whose initialization is the Laplacian eigenvector of a kNN graph derived from the oracle, updated alongside the main encoder and supervised by a cosine similarity loss. The paper reports that SaGIF has the best average ranking on NDCG@10 and competitive rankings on ERR@10 and AUC across six datasets.

Load-bearing premise

The load-bearing premise is that the feature-only cosine oracle used to measure NDCG@10 and ERR@10 is the correct ground truth for individual fairness, even though SaGIF is deliberately trained to match a different fused oracle; if the fused oracle is the right standard, the reported fairness numbers do not measure what the method optimizes.

Editorial extensions

If this is right

  • Individual fairness in GNNs can be improved by injecting a fused feature-and-topology similarity signal at the representation level, so future fair-GNN designs do not have to choose between feature-only and structure-only definitions of similarity.
  • Because the similarity encoder is initialized from a kNN graph and trained with a similarity loss, the method works with any message-passing backbone, including shallow SGC models, and adds only linear cost in the number of edges.
  • The similarity consistency diagnostic can be used to locate the nodes where individual fairness is most at risk, since the paper's experiments tie low consistency to low NDCG@10.
  • The topology-fusion and feature-fusion oracles give an alternative ground truth for 'similar individuals' that jointly uses attributes and structure, which could replace single-source similarity lists in future individual-fairness evaluations.
  • Since the fused oracle constructions do not assume homophily, the approach is claimed to transfer to graphs where feature-similar nodes are structurally far apart.

Reading between the lines

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

  • Editorial inference: the reported fairness gains are evaluated against a feature-only cosine oracle, while SaGIF is optimized against a fused oracle; if the fused oracle is the better ground truth, the current numbers may be a conservative or mismatched measurement, and validating the oracle itself would be a natural next step.
  • Editorial inference: the paper's correlation between similarity consistency and NDCG@10 is observational and may be confounded with degree, label homophily, or feature quality; a direct test is to rewire graph edges to raise consistency and see whether individual fairness rises.
  • Editorial inference: the independent similarity channel is a general inductive bias, not a fairness-specific trick; it could plausibly be reused for tasks like link prediction or recommendation, and for stabilizing node representations when topology and features are misaligned.
  • Editorial inference: the choice between Laplacian-eigenvector and random-walk initialization has a robustness implication: eigenvector signs are ambiguous, so deployments should fix a deterministic sign convention or use the walk variant when reproducibility across training runs matters.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 3 minor

Summary. The paper argues that individual unfairness in graph neural networks stems from similarity inconsistency between feature-based and topology-based similarity, proposes two oracle similarity metrics (topology fusion and feature fusion), and introduces SaGIF, a method that learns independent similarity representations initialized from the fused oracle and trained with a similarity-preservation loss. The experimental study compares SaGIF against PFR, InFoRM, and REDRESS on six benchmark graphs using GCN and SGC backbones, reporting AUC, NDCG@10, and ERR@10, along with ablations and hyperparameter analyses.

Significance. If the empirical claims were fully supported, SaGIF would be a useful contribution: the idea of decoupling similarity encoding from the task encoder is clean, the method is compatible with multiple GNN backbones, the authors provide code, and the ablation study isolates the contribution of the similarity encoder, the kNN graph, and the similarity loss. However, the central evaluation currently measures individual fairness with a feature-only cosine oracle while SaGIF is trained to preserve a fused topology-and-feature oracle, so the reported NDCG/ERR gains do not yet validate the method's intended fairness objective. The abstract also overstates the consistency of the empirical wins relative to the tables.

major comments (4)
  1. [VI-A2, Eq. (12), Eqs. (4)-(7)] The fairness evaluation uses a feature-cosine oracle, while SaGIF's similarity loss in Eq. (12) regresses output similarity toward the fused oracle S defined by topology fusion or feature fusion in Eqs. (4)-(7). Section VI-A2 states that all NDCG@10 and ERR@10 oracle matrices are 'calculated using feature similarity determined by cosine similarity' for consistency with previous studies, but this means the reported fairness gains do not measure whether SaGIF preserves the fused similarity metric that motivates the method. A model could rank nodes by feature-cosine similarity and score well on the evaluation while being unfair under the fused oracle, or vice versa. The authors should report NDCG@10 and ERR@10 under both the feature-cosine oracle and the fused oracle, or otherwise justify why the feature-only oracle is the correct ground truth; without this, the central claim that SaGIF improves individual fairness is not established.
  2. [Abstract, Tables II and III] The abstract claims that SaGIF 'consistently outperforms state-of-the-art IF methods,' but Tables II and III show multiple cases where REDRESS achieves the best NDCG@10: Cora-GCN (60.00 vs. 58.88), Coauthor-phy-GCN (37.35 vs. 32.28), Pubmed-SGC (42.26 vs. 40.01), and Coauthor-phy-SGC (52.85 vs. 50.15), with additional ERR@10 wins as well. The average-ranking summary in Table IV supports a softer claim that SaGIF is competitive and often best, but the 'consistently outperforms' wording is contradicted by the detailed tables. The abstract and conclusion should be revised to reflect the actual per-dataset outcomes, such as 'improves IF in most cases' or 'achieves the best average ranking.'
  3. [III-B, Figure 1] The paper states that 'the source of individual unfairness in GNNs can be traced back to the issue of similarity inconsistency' and that low similarity consistency harms individual fairness, but Figure 1 only shows binned correlations between the proposed similarity-consistency measure and NDCG@10. The accompanying message-passing explanation is qualitative, and no intervention varies similarity consistency while controlling for confounders, so the causal reading is not justified. The authors should soften the causal language or provide controlled evidence, for example by constructing synthetic graphs with matched feature distributions and systematically varied feature-topology consistency.
  4. [V-D, Eq. (10), Section IV-B] The contribution statement and Section V-D claim that SaGIF 'incurs only a linear complexity cost,' but the complexity analysis only covers the message-passing layers. The initialization pipeline includes all-pairs Dijkstra for feature fusion in Section IV-B, kNN graph construction from the oracle similarity matrix, and Laplacian eigendecomposition in Eq. (10), which are generally O(n^2) or worse. The complexity claim should be restricted to per-epoch training after initialization, or the preprocessing costs should be analyzed explicitly and included in the stated complexity.
minor comments (3)
  1. [VI-E1] In the discussion of small d_sim, the text says 'the downstream task loss function L_s fails to converge,' but L_s is the similarity loss, not the downstream task loss; this appears to be a typo and should be corrected to L_t or rephrased.
  2. [Figure 2] Figure 2 contains many labels that appear unrelated to the described similarity metrics (for example, 'Demo Neighbor Assigner,' 'Channel Discriminator,' and 'Disentangled Layers'), which is likely a figure-generation artifact; the figure should be regenerated so that its annotations match the topology-fusion and feature-fusion descriptions.
  3. [VI-B, Table IV] The comparison study reports means and standard deviations but no statistical significance tests; adding paired significance tests across the five runs would make the average-ranking claims in Table IV more conclusive, especially where differences are small (e.g., Cora-GCN NDCG@10).

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: SaGIF's fused similarity encoder is trained against its own S, but the reported IF gains are measured against an external feature-cosine oracle, so the central claim is not forced by construction.

full rationale

The derivation chain does not reduce to its inputs. SaGIF's similarity encoder is initialized from Laplacian eigenvectors of a kNN graph built from its own fused oracle S (Eq. 10 and Algorithm 1 lines 2-4) and trained with the similarity loss L_s(P,S) = ||D(P,P^T) - S||_F^2 / (k|V_train|) (Eq. 12), so the statement that the encoder preserves the fused oracle is true by design. However, the paper does not present that internal preservation as the measured fairness result. Section VI-A2 explicitly states that all NDCG@10 and ERR@10 evaluation oracles are calculated using feature-cosine similarity, not the fused S that SaGIF is trained to preserve; therefore Tables II and III compare SaGIF against an external benchmark. A model could perfectly preserve fused S and still fare arbitrarily on feature-cosine NDCG, so the reported IF improvements are empirical rather than forced. The preliminary similarity-consistency correlation in Section III-B and Figure 1 is only correlational; the causal claim that similarity inconsistency is 'the source' of individual unfairness is an interpretive overreach, but no equation in the paper reduces that conclusion to its own definition. The only self-citation identified is reference [24] (RawlsGCN, sharing first author Yuchang Zhu) in the related-work taxonomy of degree-related fairness; it is not load-bearing. No uniqueness theorem is imported from the authors' prior work, and the positional-encoding machinery is attributed to external prior work [42]. The odd sentence in Section VI-E1 saying 'the downstream task loss function L_s fails to converge' appears to be a typo for L_t, but it is not part of the circularity chain. Overall, no Eq. X = Eq. Y reduction or fitted-parameter-renamed-as-prediction is present, so the score reflects only a minor, non-load-bearing self-citation and no significant circularity.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claims rest on five domain assumptions and four fitted hyperparameters; no new physical or formal entities are introduced. The fused oracle definitions depend on k and lambda, and the reported results depend on d_sim and alpha, so the method's wins are not parameter-free.

free parameters (4)
  • k (top-k neighborhood size) = 10
    Used to compute similarity consistency (Eq. 3), to build the kNN graph in Eq. (4), to construct the similarity-encoding kNN graph, and as the ranking cutoff for NDCG@k and ERR@k. No sensitivity analysis is reported for k, so behavior away from k=10 is unknown.
  • lambda = 0.5
    In Eq. (4), lambda balances normalized adjacency (structure) and normalized kNN graph (features) in topology fusion. Fixed at 0.5 without tuning or ablating, though it directly controls the fused oracle.
  • d_sim = searched over {8, 16, 32, 64, 128}
    Dimension of the initial similarity encoding P0 (Eq. 10) and of the similarity representations. The paper shows fairness improves with d_sim, so the results depend on this choice.
  • alpha = searched over 0 to 1; sensitivity up to 100
    Weight of the similarity loss L_s in Eq. (11). The search protocol is not described as validation-based, and the sensitivity analysis shows performance declines for large alpha.
assumptions (5)
  • domain assumption Similarity consistency, as defined by top-k overlap of feature-based and topology-based neighbor lists (Def. 2), is a meaningful measure of how well GNNs can identify similar individuals.
    The paper's motivational analysis and causal interpretation rest on this quantity, which has no external validation.
  • ad hoc to paper The binned correlation in Figure 1(b)-(d) reflects a causal relationship: low similarity consistency causes individual unfairness in GNNs.
    Section III-B interprets the correlation as the source of unfairness, then uses it to justify the fused oracle. No experiment manipulates consistency to confirm causation.
  • domain assumption Feature-cosine similarity is the correct oracle for measuring individual fairness in the evaluation.
    Section VI-A2 fixes all fairness metrics to feature-only cosine similarity for comparability, even though the method's own oracle is fused. If the fused oracle is the right standard, the reported NDCG and ERR values do not reflect it.
  • domain assumption Laplacian eigenvectors of the kNN graph built from S preserve the oracle similarity information.
    Eq. (10) assumes the top d_sim eigenvectors of L_k provide a useful initial encoding; the paper relies on this for the main effect but does not prove or measure the preservation.
  • domain assumption GNN message passing over graph edges is the mechanism through which similarity inconsistency translates to unfair outcomes.
    Section III-B gives an informal aggregation argument; this is not formalized or tested against alternative mechanisms.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SaGIF: Improving Individual Fairness in Graph Neural Networks via Similarity Encoding." pith.science (2026). https://pith.science/paper/2TNFWFK3

@misc{pith2026250618696,
  author       = {Pith},
  title        = {Pith review of: SaGIF: Improving Individual Fairness in Graph Neural Networks via Similarity Encoding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2TNFWFK3}},
  note         = {Machine review of arXiv:2506.18696}
}
read the original abstract

Individual fairness (IF) in graph neural networks (GNNs), which emphasizes the need for similar individuals should receive similar outcomes from GNNs, has been a critical issue. Despite its importance, research in this area has been largely unexplored in terms of (1) a clear understanding of what induces individual unfairness in GNNs and (2) a comprehensive consideration of identifying similar individuals. To bridge these gaps, we conduct a preliminary analysis to explore the underlying reason for individual unfairness and observe correlations between IF and similarity consistency, a concept introduced to evaluate the discrepancy in identifying similar individuals based on graph structure versus node features. Inspired by our observations, we introduce two metrics to assess individual similarity from two distinct perspectives: topology fusion and feature fusion. Building upon these metrics, we propose Similarity-aware GNNs for Individual Fairness, named SaGIF. The key insight behind SaGIF is the integration of individual similarities by independently learning similarity representations, leading to an improvement of IF in GNNs. Our experiments on several real-world datasets validate the effectiveness of our proposed metrics and SaGIF. Specifically, SaGIF consistently outperforms state-of-the-art IF methods while maintaining utility performance. Code is available at: https://github.com/ZzoomD/SaGIF.

Figures

Figures reproduced from arXiv: 2506.18696 by the authors.

Figure 1
Figure 1. The preliminary analysis on three datasets. (a) Similarity consistency statistic of three datasets. (b)-(d) Node classification results, [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The illustration of similarity measurement from two perspec [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. The overview of our proposed method SaGIF. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Hyperparameter analysis w.r.t. the balanced hyperparameter [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Impact analysis w.r.t. similarity encoding dimensions [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

48 extracted references · 32 canonical work pages

  1. [1]

    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

    arXiv 2016

  2. [2]

    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

  3. [3]

    Mod- eling two-way selection preference for person-job fit,

    C. Yang, Y . Hou, Y . Song, T. Zhang, J.-R. Wen, and W. X. Zhao, “Mod- eling two-way selection preference for person-job fit,” inProceedings of the 16th ACM Conference on Recommender Systems, 2022, pp. 102–112

    work page 2022

  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,” inThe world wide web conference, 2019, pp. 417–426

    2019

  5. [5]

    An effective self-supervised framework for learning expressive molecular global representations to drug discovery,

    P. Li, J. Wang, Y . Qiao, H. Chen, Y . Yu, X. Yao, P. Gao, G. Xie, and S. Song, “An effective self-supervised framework for learning expressive molecular global representations to drug discovery,”Briefings in Bioin- formatics, vol. 22, no. 6, p. bbab109, 2021

    work page 2021

  6. [6]

    Say no to the discrimination: Learning fair graph neural networks with limited sensitive attribute information,

    E. Dai and S. Wang, “Say no to the discrimination: Learning fair graph neural networks with limited sensitive attribute information,” in Proceedings of the 14th ACM International Conference on Web Search and Data Mining, 2021, pp. 680–688

    work page 2021

  7. [7]

    Edits: Modeling and mitigating data bias for graph neural networks,

    Y . Dong, N. Liu, B. Jalaian, and J. Li, “Edits: Modeling and mitigating data bias for graph neural networks,” inProceedings of the ACM Web Conference 2022, 2022, pp. 1259–1269

    work page 2022

  8. [8]

    Towards a unified framework for fair and stable graph representation learning,

    C. Agarwal, H. Lakkaraju, and M. Zitnik, “Towards a unified framework for fair and stable graph representation learning,” inUncertainty in Artificial Intelligence. PMLR, 2021, pp. 2114–2124

    work page 2021

Show all 48 references
  1. [9]

    Inform: Individual fairness on graph mining,

    J. Kang, J. He, R. Maciejewski, and H. Tong, “Inform: Individual fairness on graph mining,” inProceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining, 2020, pp. 379–389

    2020

  2. [10]

    Equality of opportunity in supervised learning,

    M. Hardt, E. Price, and N. Srebro, “Equality of opportunity in supervised learning,”Advances in neural information processing systems, vol. 29, 2016

    2016

  3. [11]

    Fairness through awareness,

    C. Dwork, M. Hardt, T. Pitassi, O. Reingold, and R. Zemel, “Fairness through awareness,” inProceedings of the 3rd innovations in theoretical computer science conference, 2012, pp. 214–226

    2012

  4. [12]

    Operationalizing in- dividual fairness with pairwise fair representations,

    P. Lahoti, K. P. Gummadi, and G. Weikum, “Operationalizing in- dividual fairness with pairwise fair representations,”arXiv preprint arXiv:1907.01439, 2019

    1907 arXiv

  5. [13]

    Individual fairness for graph neural networks: A ranking based approach,

    Y . Dong, J. Kang, H. Tong, and J. Li, “Individual fairness for graph neural networks: A ranking based approach,” inProceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, 2021, pp. 300–310

    2021

  6. [14]

    Guide: Group equality informed individual fairness in graph neural networks,

    W. Song, Y . Dong, N. Liu, and J. Li, “Guide: Group equality informed individual fairness in graph neural networks,” inProceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2022, pp. 1625–1634

    2022

  7. [15]

    Gfairhint: Improving individual fairness for graph neural networks via fairness hint,

    P. Xu, Y . Zhou, B. An, W. Ai, and F. Huang, “Gfairhint: Improving individual fairness for graph neural networks via fairness hint,”arXiv preprint arXiv:2305.15622, 2023

    2023 arXiv

  8. [16]

    A survey on bias and fairness in machine learning,

    N. Mehrabi, F. Morstatter, N. Saxena, K. Lerman, and A. Galstyan, “A survey on bias and fairness in machine learning,”ACM computing surveys (CSUR), vol. 54, no. 6, pp. 1–35, 2021

    2021

  9. [17]

    Counterfactual fairness,

    M. J. Kusner, J. Loftus, C. Russell, and R. Silva, “Counterfactual fairness,”Advances in neural information processing systems, vol. 30, 2017

    2017

  10. [18]

    Learning adversarially fair and transferable representations,

    D. Madras, E. Creager, T. Pitassi, and R. Zemel, “Learning adversarially fair and transferable representations,” inInternational Conference on Machine Learning. PMLR, 2018, pp. 3384–3393

    2018

  11. [19]

    Fair representation learning for heterogeneous information networks,

    Z. Zeng, R. Islam, K. N. Keya, J. Foulds, Y . Song, and S. Pan, “Fair representation learning for heterogeneous information networks,” inProceedings of the International AAAI Conference on Web and Social Media, vol. 15, 2021, pp. 877–887

    2021

  12. [20]

    Fairer: fairness as decision rationale alignment,

    T. Li, Q. Guo, A. Liu, M. Du, Z. Li, and Y . Liu, “Fairer: fairness as decision rationale alignment,” inInternational Conference on Machine Learning. PMLR, 2023, pp. 19 471–19 489

    2023

  13. [21]

    Rawls,Justice as fairness: A restatement

    J. Rawls,Justice as fairness: A restatement. Harvard University Press, 2001

    2001

  14. [22]

    Fairness without demographics in repeated loss minimization,

    T. Hashimoto, M. Srivastava, H. Namkoong, and P. Liang, “Fairness without demographics in repeated loss minimization,” inInternational Conference on Machine Learning. PMLR, 2018, pp. 1929–1938

    2018

  15. [23]

    Learning for counterfactual fairness from observational data,

    J. Ma, R. Guo, A. Zhang, and J. Li, “Learning for counterfactual fairness from observational data,” inProceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2023, pp. 1620– 1630

    2023

  16. [24]

    Rawlsgcn: Towards rawl- sian difference principle on graph convolutional network,

    J. Kang, Y . Zhu, Y . Xia, J. Luo, and H. Tong, “Rawlsgcn: Towards rawl- sian difference principle on graph convolutional network,” inProceedings of the ACM Web Conference 2022, 2022, pp. 1214–1225

    2022

  17. [25]

    Learning fair representations,

    R. Zemel, Y . Wu, K. Swersky, T. Pitassi, and C. Dwork, “Learning fair representations,” inInternational conference on machine learning. PMLR, 2013, pp. 325–333

    2013

  18. [26]

    ifair: Learning individually fair data representations for algorithmic decision making,

    P. Lahoti, K. P. Gummadi, and G. Weikum, “ifair: Learning individually fair data representations for algorithmic decision making,” in2019 ieee 35th international conference on data engineering (icde). IEEE, 2019, pp. 1334–1345

    2019

  19. [27]

    Accurate fairness: Improving individual fairness without trading accuracy,

    X. Li, P. Wu, and J. Su, “Accurate fairness: Improving individual fairness without trading accuracy,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 12, 2023, pp. 14 312–14 320

    2023

  20. [28]

    Two simple ways to learn individual fairness metrics from data,

    D. Mukherjee, M. Yurochkin, M. Banerjee, and Y . Sun, “Two simple ways to learn individual fairness metrics from data,” inInternational Conference on Machine Learning. PMLR, 2020, pp. 7097–7107

    2020

  21. [29]

    Verifying individual fairness in machine learning models,

    P. G. John, D. Vijaykeerthy, and D. Saha, “Verifying individual fairness in machine learning models,” inConference on Uncertainty in Artificial Intelligence. PMLR, 2020, pp. 749–758

    2020

  22. [30]

    Individual fairness for local private graph neural network,

    X. Wang, T. Gu, X. Bao, L. Chang, and L. Li, “Individual fairness for local private graph neural network,”Knowledge-Based Systems, vol. 268, p. 110490, 2023

    2023

  23. [31]

    Arnetminer: extraction and mining of academic social networks,

    J. Tang, J. Zhang, L. Yao, J. Li, L. Zhang, and Z. Su, “Arnetminer: extraction and mining of academic social networks,” inProceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, 2008, pp. 990–998

    2008

  24. [32]

    Pitfalls of graph neural network evaluation,

    O. Shchur, M. Mumme, A. Bojchevski, and S. G ¨unnemann, “Pitfalls of graph neural network evaluation,”arXiv preprint arXiv:1811.05868, 2018

    2018 arXiv

  25. [33]

    Simplifying graph convolutional networks,

    F. Wu, A. Souza, T. Zhang, C. Fifty, T. Yu, and K. Weinberger, “Simplifying graph convolutional networks,” inInternational conference on machine learning. PMLR, 2019, pp. 6861–6871

    2019

  26. [34]

    Cumulated gain-based evaluation of ir techniques,

    K. J ¨arvelin and J. Kek ¨al¨ainen, “Cumulated gain-based evaluation of ir techniques,”ACM Transactions on Information Systems (TOIS), vol. 20, no. 4, pp. 422–446, 2002

    2002

  27. [35]

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

    W.-L. Chiang, X. Liu, S. Si, Y . Li, S. Bengio, and C.-J. Hsieh, “Cluster- gcn: An efficient algorithm for training deep and large graph convolutional networks,” inProceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, 2019, pp. 257–266

    2019

  28. [36]

    Unifying graph convolutional neural net- works and label propagation,

    H. Wang and J. Leskovec, “Unifying graph convolutional neural net- works and label propagation,”arXiv preprint arXiv:2002.06755, 2020

    2002 arXiv

  29. [37]

    Attributed graph models: Modeling network structure with correlated attributes,

    J. J. Pfeiffer III, S. Moreno, T. La Fond, J. Neville, and B. Gallagher, “Attributed graph models: Modeling network structure with correlated attributes,” inProceedings of the 23rd international conference on World wide web, 2014, pp. 831–842

    2014

  30. [38]

    Node similarity preserving graph convolutional networks,

    W. Jin, T. Derr, Y . Wang, Y . Ma, Z. Liu, and J. Tang, “Node similarity preserving graph convolutional networks,” inProceedings of the 14th ACM international conference on web search and data mining, 2021, pp. 148–156

    2021

  31. [39]

    Position-aware subgraph neural networks with data-efficient learning,

    C. Liu, Y . Yang, Z. Xie, H. Lu, and Y . Ding, “Position-aware subgraph neural networks with data-efficient learning,” inProceedings of the Six- teenth ACM International Conference on Web Search and Data Mining, 2023, pp. 643–651

    2023

  32. [40]

    Position-aware graph neural net- works,

    J. You, R. Ying, and J. Leskovec, “Position-aware graph neural net- works,” inInternational conference on machine learning. PMLR, 2019, pp. 7134–7143

    2019

  33. [41]

    A note on two problems in connexion with graphs,

    E. W. Dijkstra, “A note on two problems in connexion with graphs,” in Edsger Wybe Dijkstra: His Life, Work, and Legacy, 2022, pp. 287–290

    2022

  34. [42]

    Graph neural networks with learnable structural and positional representations,

    V . P. Dwivedi, A. T. Luu, T. Laurent, Y . Bengio, and X. Bresson, “Graph neural networks with learnable structural and positional representations,” arXiv preprint arXiv:2110.07875, 2021

    2021 arXiv

  35. [43]

    Revisiting semi- supervised learning with graph embeddings,

    Z. Yang, W. W. Cohen, and R. Salakhutdinov, “Revisiting semi- supervised learning with graph embeddings,” 2016

    2016

  36. [44]

    Expected recipro- cal rank for graded relevance,

    O. Chapelle, D. Metlzer, Y . Zhang, and P. Grinspan, “Expected recipro- cal rank for graded relevance,” inProceedings of the 18th ACM conference on Information and knowledge management, 2009, pp. 621–630

    2009

  37. [45]

    Benchmarking graph neural networks,

    V . P. Dwivedi, C. K. Joshi, A. T. Luu, T. Laurent, Y . Bengio, and X. Bresson, “Benchmarking graph neural networks,” 2022

    2022

  38. [46]

    Graphit: Encoding graph structure in transformers,

    G. Mialon, D. Chen, M. Selosse, and J. Mairal, “Graphit: Encoding graph structure in transformers,”arXiv preprint arXiv:2106.05667, 2021

    2021 arXiv

  39. [47]

    Subgroup generalization and fairness of graph neural networks,

    J. Ma, J. Deng, and Q. Mei, “Subgroup generalization and fairness of graph neural networks,”Advances in Neural Information Processing Systems, vol. 34, pp. 1048–1061, 2021

    2021

  40. [48]

    Post-processing for individual fairness,

    F. Petersen, D. Mukherjee, Y . Sun, and M. Yurochkin, “Post-processing for individual fairness,”Advances in Neural Information Processing Systems, vol. 34, pp. 25 944–25 955, 2021

    2021

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.