Pith. sign in

REVIEW 5 major objections 6 minor 40 references

Rethinking Graph Out-Of-Distribution Generalization: A Learnable Random Walk Perspective

T0 review · 5 major / 6 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 This paper claims that learnable random walk sequences, rather than invariant topology or spectrum, are the right carrier of invariant knowledge for graph OOD generalization, and that LRW-OOD realizes this with a learnable transition…

desk verdict Sound empirical idea and a genuinely new invariant carrier for graph OOD, but Eq. (2) underspecifies the sampler and the theory is mostly restatement; worth a serious referee after fixes. read the letter →

arxiv 2505.05785 v1 pith:X3NJCX2I submitted 2025-05-09 cs.LG

Henan Sun , Xunkai Li , Lei Zhu , Junyi Han , Guang Zeng , Ronghua Li , Guoren Wang This is my paper · ORCID
classification cs.LG
keywords graphout-of-distributiongeneralizationlearnablerandomwalkmutualinformationkerneldensityestimationneuralnetworkinvariantrepresentationnodeclassificationdistributionshift
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 when graphs shift between training and test time, the invariant signal is best carried by learnable random-walk sequences rather than by invariant graph topology or spectrum. LRW-OOD samples paths with transition probabilities set by the cosine similarity between learned node embeddings, encodes each path with an MLP, and trains with a kernel-density-estimation mutual-information loss that enforces sufficiency (the label is predictable from the path embedding) and invariance (the label distribution is stable across environments). The authors prove that an encoder optimized this way satisfies both conditions and is the solution to the worst-case graph OOD objective. This matters because GNNs degrade under distribution shifts, and on seven benchmarks the paper reports an average accuracy gain of 3.87% over nine graph OOD baselines using either GCN or GAT backbones.

What carries the argument

The load-bearing object is the learnable random walk: a sampler whose transition probability from node $v_i$ to neighbor $v_j$ is $\cos(z_i,z_j)$, where $z$ is produced by a GNN backbone, so topology and node features enter the path distribution together. Each sampled path $p^r_i$ is encoded by an MLP into an LRW embedding $h^r_i$, and a KDE-based mutual-information loss (sufficiency maximization plus variance-minimizing risk extrapolation) shapes those embeddings. This replaces the fixed degree-normalized adjacency matrix used as the transition matrix in prior random-walk GNNs and is the mechanism through which the paper's theoretical guarantees and empirical gains are supposed to flow.

What would settle it

Compute the row sums of the transition matrix $[\cos(z_i,z_j)]$ over each node's neighbors on Cora (or any benchmark graph) with the trained encoder; if any row does not sum to one, the random walk sampler in Eq. (2) has no valid probability distribution and the reported results must rely on an unstated normalization or a different mechanism.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that invariant knowledge in graph OOD generalization can be instantiated as the probability of a random walk moving to a semantically similar neighbor, and that sampling $k$ such paths per node and encoding them produces representations that carry the label information while discarding environment-specific correlations. The contribution is LRW-OOD: a learnable random walk encoder whose transition matrix is parameterized by a GNN (initialized from cosine similarities), a path encoder that maps each sampled path to an embedding, and a KDE-based mutual-information loss combining MI sufficiency maximization with risk extrapolation minimization. Theorem 3.1 states that minimizing this loss yields an encoder satisfying the sufficiency and invariance conditions, and Theorem 3.2 states that such an encoder solves the worst-case formulation in Eq. (1). The experiments across synthetic, cross-domain, and temporal shift datasets support the paper's claim that the approach outperforms the compared baselines.

Load-bearing premise

The load-bearing premise is that the raw cosine similarities $\cos(z_i,z_j)$ between learned node embeddings can serve as random-walk transition probabilities, but the paper never specifies how these values are normalized or made nonnegative, so the sampler in Eq. (2) is not a well-defined probability distribution as written.

Editorial extensions

If this is right

  • If the central claim is correct, graph OOD generalization no longer requires an invariant topology or an invariant spectrum; any graph with shared semantic structure can be handled by learning transition probabilities.
  • The KDE-based mutual-information loss gives a differentiable way to push toward the sufficiency condition, replacing the KL-divergence approximations used by earlier graph OOD methods.
  • The two-stage design means the LRW encoder can be trained once and then combined with standard GCN or GAT classifiers, so the claimed gains transfer across backbone architectures.
  • The complexity bound of $O(nd^2(l_1+l_2)+nksd)$ time and $O(nd(l_1+l_2))$ space suggests the approach remains practical for larger graphs, including the temporal ogb-ArXiv benchmark.
  • Across the evaluated synthetic, cross-domain, and temporal datasets, the paper reports an average improvement of 3.87% over nine graph OOD baselines.

Reading between the lines

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

  • Editorial inference: as written, Eq. (2) uses raw cosine similarities as transition probabilities; because these values can be negative and need not sum to one over a node's neighbors, a softmax or other normalization is almost certainly required for the sampler to be well-defined, and comparing the stated version with an explicitly normalized version would clarify the actual mechanism.
  • Editorial inference: random-walk paths of bounded length are a restricted class of subgraphs, so the paper's perspective connects naturally to subgraph- and motif-based invariant learning; future work could ask which path lengths and shapes carry the invariant signal for a given shift.
  • Editorial inference: the KDE bandwidth and covariance choices in Eq. (4) are likely to affect finite-sample behavior, so a controlled comparison against histogram-based or neural mutual-information estimators on small graphs with known true mutual information would test whether the loss itself, rather than the overall architecture, drives the gains.
  • Editorial inference: path embeddings trained to be invariant could double as an OOD detection signal, since paths that remain predictive of the label across environments are invariant while environment-specific paths stand out as spurious; the paper does not explore this use.
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 argues that existing graph OOD methods that treat invariant topology or spectrum as carriers of invariant knowledge are fragile, and proposes that learnable random walk sequences instantiate invariant knowledge. It introduces LRW-OOD with a GNN-based LRW sampler that builds a cosine-similarity transition matrix, samples k random walk paths per node, encodes them with an MLP, and trains with a KDE-based MI loss combining sufficiency maximization and risk-extrapolation variance minimization. The manuscript states Theorem 3.1 (optimizing the loss yields sufficiency and invariance), Theorem 3.2 (the optimized encoder solves the worst-case OOD objective), and Theorem 3.3 (complexity bounds), and reports experiments on seven datasets with nine baselines, claiming an average 3.87% improvement.

Significance. The perspective that invariant knowledge can be carried by learnable walk distributions rather than fixed topology or spectrum is interesting and potentially useful for OOD node classification. The experimental evaluation is broad: 10-run means with standard deviations, seven datasets spanning synthetic, cross-domain, and temporal shifts, nine baselines, ablations, and hyperparameter sensitivity, and the results are consistently in favor of the proposed model. The complexity analysis is also a nice practical addition. However, the theoretical guarantees claimed in Theorems 3.1 and 3.2 are not established by the current derivations, and the transition-probability definition is incomplete, so the significance of the paper currently rests on the empirical results.

major comments (5)
  1. [Section 3.2, Eq. (2)] The transition rule in Eq. (2) does not define a valid probability distribution: cos(z_i,z_j) lies in [-1,1], need not sum to one over N(v_i), and can be negative. Because the LRW sampler, the path embeddings in Eq. (3), and the KDE-based MI losses in Eqs. (4)-(5) all depend on sampling from this transition matrix, the method and the proofs in Section 3.3 are undefined as stated. Please specify the normalization (e.g., softmax, temperature, degree-based renormalization) and how negative similarities are handled; this choice is part of the method, not an implementation detail.
  2. [Section 3.2, Eq. (4) and Appendix A, Lemma A.1] Eq. (4) defines LMI_Suff as a pointwise kernel-density ratio, but a mutual information estimate is a sum or integral over sample points of the log-ratio weighted by the joint density; the expression shown is not an MI estimate. Lemma A.1 claims convergence of 'the kernel density estimation as defined in equations 4', but no consistent estimator is actually specified. Please replace Eq. (4) with a proper KDE-based MI estimator, such as an average over the observed samples, and provide the bandwidth conditions under which Lemma A.1 holds.
  3. [Appendix A.1, proof of Theorem 3.1] The proof of Theorem 3.1 uses the claim 'I(y,e) ≤ I(y,e|h)' at the start of the invariance argument. The identity I(Y;E) = I(Y;E|H) + I(Y;H) - I(Y;H|E) shows this inequality is not generally true, and Eq. (12) writes I(y,e|h) as KL(Pr(y|h,e)||Pr(y|h)), which omits the expectation over h and the correct conditional distribution of e. Consequently Theorem 3.1 is not proven as stated.
  4. [Section 3.3, Theorem 3.1] Theorem 3.1 largely restates the construction: the loss in Eq. (5) is deliberately composed of an MI-sufficiency term and a risk-extrapolation variance term, so 'if the loss is optimized then the encoder satisfies sufficiency and invariance' is close to a tautology unless the proof establishes that the optima of these terms truly correspond to the conditions y=f*(G_e)+σ and Pr(y|e)=Pr(y). As it stands, the theorem adds little beyond the definition and the questionable derivations above.
  5. [Appendix A.2, proof of Theorem 3.2] The proof of Theorem 3.2 assumes that sufficiency implies the loss of f* is minimized for any graph G'_e and any comparator f, and that invariance implies 'the loss of f* is minimized under any graph with arbitrary environment'; neither implication follows from the stated conditions. The jump from sufficiency and invariance to the worst-case objective in Eq. (1) requires additional assumptions about the class of environments and the loss, which are not given.
minor comments (6)
  1. [Table 1 caption] The caption contains the typo 'corss-domain' and should read 'cross-domain'; also, the very large standard deviations for DGNN and MARIO on WebKB suggest that the comparison could benefit from a more careful presentation.
  2. [Appendix A.1] There is a typo 'chain low of mutual information' that should be 'chain rule of mutual information', and 'Jenson Inequality' should be 'Jensen's inequality'.
  3. [Appendix A, Lemma A.1 and Eq. (6)] The lemma states that the KDE estimator 'converges in probability' but Eq. (6) asserts almost-sure convergence; additionally, the proof cites Glivenko-Cantelli for the empirical CDF, which is not directly a pointwise KDE density convergence result. Please clarify the mode of convergence and the required assumptions on the bandwidth.
  4. [Section 4.1] The text says 'each experiment is repeated 10 times, and the average performance is reported,' but the tables report mean ± std; please clarify how the standard deviation is computed and how random seeds are handled.
  5. [Abstract and Section 4.2] The abstract claims an average improvement of 3.87% over state-of-the-art baselines, while Section 4.2 reports about 3.1% (GCN) and 1.0% (GAT) on Table 1; please state where the 3.87% figure is computed.
  6. [Reproducibility] No code is released, and the unstated normalization in Eq. (2) makes the method hard to reproduce; please release the implementation or provide all implementation details.

Circularity Check

2 steps flagged · score 6.0 of 10

Theorem 3.1 restates its own input: Eq. (5)'s KDE-MI loss is explicitly built from the sufficiency and invariance conditions it is then 'proved' to enforce, and Theorem 3.2 only re-imports Wu et al.'s equivalence, so the OOD guarantee is a designed-in restatement; the empirical part is independent.

  1. self definitional [Sec. 3.2 'KDE-based MI Loss Calculator' (Eqs. 4-5); Sec. 3.3 Theorem 3.1]
    "To encourage the LRW encoder to generate node embeddings that capture invariant features while discarding spurious correlations, we introduce two complementary objectives: the MI sufficiency maximization loss and the risk extrapolation minimization loss, which correspond to the sufficiency and invariance conditions, respectively. ... Theorem 3.1. Let f(Ge) denotes the learnable random walk encoder."

    The two OOD conditions are imported in Sec. 3.1 as 'theoretically equivalent' to Eq. (1). Eq. (5) is then assembled from exactly those two conditions: L_MI_Suff (Eq. 4) is negative pointwise KDE mutual information (sufficiency: max_f I(y,f)), and the variance V({L_MI_Suff}) is named 'risk extrapolation minimization' (invariance: min_f I(y,e|f)). Theorem 3.1 asserts that optimizing this loss yields encoders satisfying both conditions, and the proof's only mechanism is that 'minimizing LMI_Suff...

  2. other [Sec. 3.1 'Motivation' (imported equivalence) and Sec. 3.3 Theorem 3.2]
    "Wu et al. [29] introduces two conditions that are theoretically equivalent to the aforementioned graph OOD generalization formulation while also being directly applicable to graph OOD learning. ... Theorem 3.2. Let f*(Ge) denotes the optimized learnable random walk encoder satisfying both the sufficiency and invariance conditions. Then, the encoder f*(Ge) is the solution to the graph OOD generalization formulated as Equation 1."

    Theorem 3.2 is presented as the paper's own theoretical result, but its premise ('satisfying both the sufficiency and invariance conditions') is precisely what Sec. 3.1 granted as 'theoretically equivalent' to Eq. (1) via Wu et al. [29]. The proof in Appendix A.2 does not re-derive this equivalence; it only applies it under the two conditions, so the 'solution to the graph OOD generalization formulated as Equation 1' conclusion is the imported equivalence restated. Combined with Theorem 3.1 being a restatement of the loss design, the claimed guarantee chain 'loss -> conditions -> Eq. (1)' contains no independent derivation step.

full rationale

The derivation chain is: Eq. (1) (min-max OOD objective) -> two OOD conditions imported verbatim from Wu et al. [29] in Sec. 3.1 -> a loss (Eqs. 4-5) constructed from exactly those two conditions ('MI sufficiency maximization loss' = negative KDE mutual information; 'risk extrapolation minimization loss' = variance across walks) -> Theorem 3.1 'proving' the optimized encoder satisfies those conditions -> Theorem 3.2 declaring it the solution of Eq. (1). The only genuinely derived link is the KDE convergence lemma (Lemma A.1), which is standard. The sufficiency and invariance guarantees are restatements of the loss design rather than independent derivations: minimizing L_MI_Suff is, by Eq. (4)'s definition, maximizing I(h,y), and the variance term is named for the invariance condition it is said to realize. Theorem 3.2 then assumes the two conditions as premises, so its content is the equivalence already granted to Wu et al., not a new proof. Thus the theory section partially reduces to its own construction. However, the paper's headline empirical claim (3.87% over nine baselines on seven datasets) is independently testable, the OOD equivalence is an external citation rather than a self-citation chain, and the paper's self-citations ([13, 23, 22]) concern unrelated node-classification techniques, not the OOD theory. Two adjacent weaknesses are correctness issues, not circularity: Eq. (2)'s raw cosine similarities need not be non-negative or sum to one, and Lemma A.1/Theorem 3.1 uses pointwise density-ratio terms where a summed estimator would be needed; these do not affect the circularity score. Verdict: partial circularity in the theoretical guarantee (6), with the empirical contribution independent.

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

The strongest theoretical result depends on assumptions pulled from citations or introduced ad hoc: the EERM condition equivalence, a valid-but-unspecified transition probability normalization, and KDE-based MI as a faithful objective. These are not derived or externally verified here. No new physical entities are introduced.

free parameters (3)
  • random walk length s = per dataset, e.g., 1 on Cora, 5 on LastFMAsia
    Table 4 shows s is tuned per dataset and affects accuracy by up to several points.
  • number of random walks k = per dataset, generally 3 to 5
    Table 5 shows k tuned per dataset; the paper states multiple walks help.
  • KDE bandwidth m = unspecified
    Eq. (4) depends on kernel bandwidth m, but the paper does not state how it is set; it directly controls the MI estimate.
assumptions (4)
  • domain assumption The graph OOD objective Eq. (1) is equivalent to sufficiency and invariance conditions from Wu et al.
    Invoked in Section 3.1 and used to justify the loss design; no proof is given in this paper, only a citation to EERM.
  • ad hoc to paper Cosine similarity between learned node embeddings can serve as a transition probability for random walk sampling
    Eq. (2) sets Pr(vi -> vj) = cos(z_i, z_j) without normalization or sign handling; this is an unstated modeling assumption.
  • standard math Kernel density estimation with a Gaussian kernel converges to the true mutual information and is optimizable as a loss
    Lemma A.1 asserts convergence via Glivenko-Cantelli, but the paper does not verify bandwidth conditions and Eq. (4) is a pointwise density product, not a sum over samples.
  • domain assumption Invariant knowledge can be encoded in the probability of switching to the next node with similar semantic information
    Motivation M1 in Section 3.1 states this but does not prove it; it is the conceptual bridge from topology/spectrum to random walks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Rethinking Graph Out-Of-Distribution Generalization: A Learnable Random Walk Perspective." pith.science (2026). https://pith.science/paper/X3NJCX2I

@misc{pith2026250505785,
  author       = {Pith},
  title        = {Pith review of: Rethinking Graph Out-Of-Distribution Generalization: A Learnable Random Walk Perspective},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/X3NJCX2I}},
  note         = {Machine review of arXiv:2505.05785}
}
read the original abstract

Out-Of-Distribution (OOD) generalization has gained increasing attentions for machine learning on graphs, as graph neural networks (GNNs) often exhibit performance degradation under distribution shifts. Existing graph OOD methods tend to follow the basic ideas of invariant risk minimization and structural causal models, interpreting the invariant knowledge across datasets under various distribution shifts as graph topology or graph spectrum. However, these interpretations may be inconsistent with real-world scenarios, as neither invariant topology nor spectrum is assured. In this paper, we advocate the learnable random walk (LRW) perspective as the instantiation of invariant knowledge, and propose LRW-OOD to realize graph OOD generalization learning. Instead of employing fixed probability transition matrix (i.e., degree-normalized adjacency matrix), we parameterize the transition matrix with an LRW-sampler and a path encoder. Furthermore, we propose the kernel density estimation (KDE)-based mutual information (MI) loss to generate random walk sequences that adhere to OOD principles. Extensive experiment demonstrates that our model can effectively enhance graph OOD generalization under various types of distribution shifts and yield a significant accuracy improvement of 3.87% over state-of-the-art graph OOD generalization baselines.

Figures

Figures reproduced from arXiv: 2505.05785 by the authors.

Figure 1
Figure 1. An example of the heterophilic citation network under temporal distribution shift and pipelines of [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The framework of the proposed LRW-OOD. min f max Ge∼G L(f(Ge), y), (1) where G is the set of graphs under all environments, Ge is the graph under the environment e, L is the loss function, f is the graph OOD model and y is the label. However, this optimization formulation cannot be directly applied to graph OOD generalization due to the inaccessibility of environmental factors e. To address this challenge, Wu et al.… view at source ↗
Figure 3
Figure 3. The performance comparison of graph OOD models using GCN as the backbone. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The performance comparison of graph OOD models using GAT as the backbone. [PITH_FULL_IMAGE:figures/full_fig_p018_4.png]
Figure 5
Figure 5. Figure 5: The visualization of weights of LRW-OOD on the synthetic datasets. [PITH_FULL_IMAGE:figures/full_fig_p019_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 28 canonical work pages

  1. [1]

    Invariant risk mini- mization

    Martin Arjovsky, Léon Bottou, Ishaan Gulrajani, and David Lopez-Paz. Invariant risk mini- mization. arXiv preprint arXiv:1907.02893, 2019

    arXiv 1907

  2. [2]

    Probability and measure theory

    Robert B Ash and Catherine A Doléans-Dade. Probability and measure theory. Academic press, 2000

    work page 2000

  3. [3]

    Graph neural networks in network neuroscience

    Alaa Bessadok, Mohamed Ali Mahjoub, and Islem Rekik. Graph neural networks in network neuroscience. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(5):5833– 5848, 2022

    work page 2022

  4. [4]

    Gccad: Graph contrastive learning for anomaly detection

    Bo Chen, Jing Zhang, Xiaokang Zhang, Yuxiao Dong, Jian Song, Peng Zhang, Kaibo Xu, Evgeny Kharlamov, and Jie Tang. Gccad: Graph contrastive learning for anomaly detection. IEEE Transactions on Knowledge and Data Engineering, 2022

    work page 2022

  5. [5]

    Learning causally invariant representations for out-of- distribution generalization on graphs

    Yongqiang Chen, Yonggang Zhang, Yatao Bian, Han Yang, MA Kaili, Binghui Xie, Tongliang Liu, Bo Han, and James Cheng. Learning causally invariant representations for out-of- distribution generalization on graphs. Advances in Neural Information Processing Systems, 35:22131–22148, 2022

    work page 2022

  6. [6]

    Investigating out-of-distribution generalization of gnns: An architecture perspective

    Kai Guo, Hongzhi Wen, Wei Jin, Yaming Guo, Jiliang Tang, and Yi Chang. Investigating out-of-distribution generalization of gnns: An architecture perspective. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 932–943, 2024

    work page 2024

  7. [7]

    Inductive representation learning on large graphs

    Will Hamilton, Zhitao Ying, and Jure Leskovec. Inductive representation learning on large graphs. Advances in Neural Information Processing Systems, NeurIPS, 2017

    work page 2017

  8. [8]

    Open graph benchmark: Datasets for machine learning on graphs

    Weihua Hu, Matthias Fey, Marinka Zitnik, Yuxiao Dong, Hongyu Ren, Bowen Liu, Michele Catasta, and Jure Leskovec. Open graph benchmark: Datasets for machine learning on graphs. Advances in Neural Information Processing Systems, NeurIPS, 2020

    work page 2020

Show all 40 references
  1. [9]

    Semi-supervised classification with graph convolutional networks

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

    2017

  2. [10]

    Klicpera, A

    J. Klicpera, A. Bojchevski, and S Günnemann. Predict then propagate: Graph neural networks meet personalized pagerank. In International Conference on Learning Representations, ICLR, 2019

    2019

  3. [11]

    Out-of-distribution generalization with maximal invariant predictor

    Masanori Koyama and Shoichiro Yamaguchi. Out-of-distribution generalization with maximal invariant predictor. 2020

    2020

  4. [12]

    Estimating mutual information

    Alexander Kraskov, Harald Stögbauer, and Peter Grassberger. Estimating mutual information. Physical Review E—Statistical, Nonlinear, and Soft Matter Physics, 69(6):066138, 2004

    2004

  5. [13]

    Rethinking node-wise propagation for large-scale graph learning

    Xunkai Li, Jingyuan Ma, Zhengyu Wu, Daohan Su, Wentao Zhang, Rong-Hua Li, and Guoren Wang. Rethinking node-wise propagation for large-scale graph learning. In Proceedings of the ACM Web Conference, WWW, 2024

    2024

  6. [14]

    Revisiting graph contrastive learning from the perspective of graph spectrum

    Nian Liu, Xiao Wang, Deyu Bo, Chuan Shi, and Jian Pei. Revisiting graph contrastive learning from the perspective of graph spectrum. Advances in Neural Information Processing Systems, 35:2972–2983, 2022

    2022

  7. [15]

    Estimation of mutual information using kernel density estimators

    Young-Il Moon, Balaji Rajagopalan, and Upmanu Lall. Estimation of mutual information using kernel density estimators. Physical Review E, 52(3):2318, 1995

    1995

  8. [16]

    Geom-gcn: Geo- metric graph convolutional networks

    Hongbin Pei, Bingzhe Wei, Kevin Chen-Chuan Chang, Yu Lei, and Bo Yang. Geom-gcn: Geo- metric graph convolutional networks. InInternational Conference on Learning Representations, ICLR, 2020

    2020

  9. [17]

    Multi-scale attributed node embedding

    Benedek Rozemberczki, Carl Allen, and Rik Sarkar. Multi-scale attributed node embedding. Journal of Complex Networks, 9(2):cnab014, 2021. 11

    2021

  10. [18]

    Characteristic functions on graphs: Birds of a feather, from statistical descriptors to parametric models

    Benedek Rozemberczki and Rik Sarkar. Characteristic functions on graphs: Birds of a feather, from statistical descriptors to parametric models. In Proceedings of the 29th ACM international conference on information & knowledge management, pages 1325–1334, 2020

    2020

  11. [19]

    Density estimation for statistics and data analysis

    Bernard W Silverman. Density estimation for statistics and data analysis. Routledge, 2018

    2018

  12. [20]

    Graph-based semi-supervised learning: A comprehensive review

    Zixing Song, Xiangli Yang, Zenglin Xu, and Irwin King. Graph-based semi-supervised learning: A comprehensive review. IEEE Transactions on Neural Networks and Learning Systems, 2022

    2022

  13. [21]

    The mu- tual information: detecting and evaluating dependencies between variables

    Ralf Steuer, Jürgen Kurths, Carsten O Daub, Janko Weise, and Joachim Selbig. The mu- tual information: detecting and evaluating dependencies between variables. Bioinformatics, 18(suppl_2):S231–S240, 2002

    2002

  14. [22]

    Dirw: Path-aware digraph learning for heterophily

    Daohan Su, Xunkai Li, Zhenjun Li, Yinping Liao, Rong-Hua Li, and Guoren Wang. Dirw: Path-aware digraph learning for heterophily. arXiv preprint arXiv:2410.10320, 2024

    2024

  15. [23]

    Breaking the entanglement of homophily and heterophily in semi-supervised node classification

    Henan Sun, Xunkai Li, Zhengyu Wu, Daohan Su, Rong-Hua Li, and Guoren Wang. Breaking the entanglement of homophily and heterophily in semi-supervised node classification. arXiv preprint arXiv:2312.04111, 2023

    2023 arXiv

  16. [24]

    Rethinking graph neural networks for anomaly detection

    Jianheng Tang, Jiajin Li, Ziqi Gao, and Jia Li. Rethinking graph neural networks for anomaly detection. In International Conference on Machine Learning, ICML, 2022

    2022

  17. [25]

    Graph attention networks

    Petar Veliˇckovi´c, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. Graph attention networks. In International Conference on Learning Representations, ICLR, 2018

    2018

  18. [26]

    Recommending related products using graph neural networks in directed graphs

    Srinivas Virinchi, Anoop Saladi, and Abhirup Mondal. Recommending related products using graph neural networks in directed graphs. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Springer, 2022

    2022

  19. [27]

    Generative risk min- imization for out-of-distribution generalization on graphs

    Song Wang, Zhen Tan, Yaochen Zhu, Chuxu Zhang, and Jundong Li. Generative risk min- imization for out-of-distribution generalization on graphs. arXiv preprint arXiv:2502.07968, 2025

    2025 arXiv

  20. [28]

    Graph out-of-distribution generalization via causal intervention

    Qitian Wu, Fan Nie, Chenxiao Yang, Tianyi Bao, and Junchi Yan. Graph out-of-distribution generalization via causal intervention. In Proceedings of the ACM Web Conference 2024, pages 850–860, 2024

    2024

  21. [29]

    Handling distribution shifts on graphs: An invariance perspective

    Qitian Wu, Hengrui Zhang, Junchi Yan, and David Wipf. Handling distribution shifts on graphs: An invariance perspective. arXiv preprint arXiv:2202.02466, 2022

    2022 arXiv

  22. [30]

    A comprehensive survey on graph neural networks

    Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and S Yu Philip. A comprehensive survey on graph neural networks. IEEE Transactions on Neural Networks and Learning Systems, 32(1):4–24, 2020

    2020

  23. [31]

    Learning invariant representations of graph neural networks via cluster generalization

    Donglin Xia, Xiao Wang, Nian Liu, and Chuan Shi. Learning invariant representations of graph neural networks via cluster generalization. Advances in Neural Information Processing Systems, 36:45602–45613, 2023

    2023

  24. [32]

    Pathmlp: Smooth path towards high-order homophily

    Chenxuan Xie, Jiajun Zhou, Shengbo Gong, Jiacheng Wan, Jiaxu Qian, Shanqing Yu, Qi Xuan, and Xiaoniu Yang. Pathmlp: Smooth path towards high-order homophily. arXiv preprint arXiv:2306.13532, 2023

    2023 arXiv

  25. [33]

    How powerful are graph neural networks? 2019

    Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. How powerful are graph neural networks? 2019

    2019

  26. [34]

    Cohen, and Ruslan Salakhutdinov

    Zhilin Yang, William W. Cohen, and Ruslan Salakhutdinov. Revisiting semi-supervised learning with graph embeddings. In International Conference on Machine Learning, ICML, 2016

    2016

  27. [35]

    Graph attention multi-layer perceptron

    Wentao Zhang, Ziqi Yin, Zeang Sheng, Yang Li, Wen Ouyang, Xiaosen Li, Yangyu Tao, Zhi Yang, and Bin Cui. Graph attention multi-layer perceptron. Proceedings of the ACM SIGKDD Conference on Knowledge Discovery and Data Mining, KDD, 2022. 12

    2022

  28. [36]

    Hierarchical graph pooling with structure learning

    Zhen Zhang, Jiajun Bu, Martin Ester, Jianfeng Zhang, Chengwei Yao, Zhi Yu, and Can Wang. Hierarchical graph pooling with structure learning. arXiv preprint arXiv:1911.05954, 2019

    1911 arXiv

  29. [37]

    Ugrec: modeling directed and undirected relations for recommendation

    Xinxiao Zhao, Zhiyong Cheng, Lei Zhu, Jiecai Zheng, and Xueqing Li. Ugrec: modeling directed and undirected relations for recommendation. InProceedings of the International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR, 2021

    2021

  30. [38]

    Graph neural networks: Taxonomy, advances, and trends

    Yu Zhou, Haixia Zheng, Xin Huang, Shufeng Hao, Dengao Li, and Jumin Zhao. Graph neural networks: Taxonomy, advances, and trends. ACM Transactions on Intelligent Systems and TechnoLoGy, 13(1):1–54, 2022

    2022

  31. [39]

    Shift-robust gnns: Overcoming the limitations of localized graph training data

    Qi Zhu, Natalia Ponomareva, Jiawei Han, and Bryan Perozzi. Shift-robust gnns: Overcoming the limitations of localized graph training data. Advances in Neural Information Processing Systems, 34:27965–27977, 2021

    2021

  32. [40]

    Mario: Model agnostic recipe for improving ood generalization of graph contrastive learning

    Yun Zhu, Haizhou Shi, Zhenshuo Zhang, and Siliang Tang. Mario: Model agnostic recipe for improving ood generalization of graph contrastive learning. In Proceedings of the ACM Web Conference 2024, pages 300–311, 2024. 13 A Theoretical Analysis for Section 3.3 Before proceeding ...

    2024

Pith tools

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