Pith. sign in

REVIEW 2 major objections 4 minor 54 references

A Recipe for Causal Graph Regression: Confounding Effects Revisited

T0 review · 2 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read Graph regression beats OOD baselines by letting confounders keep their predictive power.

desk verdict Useful empirical recipe for graph regression OOD, but the causal-intervention claim is an analogy, not a derivation. read the letter →

arxiv 2507.00440 v1 pith:ZX7YSV2I submitted 2025-07-01 cs.LG cs.AIstat.ME

classification cs.LGcs.AIstat.ME MSC 68T0762D20
keywords causalgraphregressionout-of-distributiongeneralizationinformationbottleneckconfoundingsubgraphcontrastivelearningbackdooradjustmentneuralnetworks
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

This paper argues that existing causal graph learning methods fail on regression tasks because they assume confounding subgraphs contain no predictive information, an assumption that rarely holds in practice. It proposes a framework that lets the confounder contribute to prediction through an enhanced graph information bottleneck objective, and replaces label-dependent intervention with a contrastive alignment loss that does not need discrete classes. If the framework works as claimed, it gives graph regression models a practical route to out-of-distribution generalization without discarding useful signal. The authors report that their method achieves the best out-of-distribution error on all eight GOOD-ZINC settings and on six of ten ReactionOOD settings.

What carries the argument

The load-bearing objects are the enhanced graph information bottleneck objective (Eq. 4), which adds a confounder-predictiveness term $-\beta I(S;Y)$ to the classical GIB loss, and the contrastive intervention loss (Eq. 17), which replaces label-conditioned counterfactual mixing with unsupervised InfoNCE alignment between $H_{g,i}$ and $H_{c,i}+H_{s,j}$. These are computed with variational Gaussian bounds, reducing the mutual-information terms to least-squares goals and an $\ell^2$ penalty on the causal embedding mean.

What would settle it

Run the method on a synthetic graph regression dataset with a known generative model where the confounder $S$ has no causal effect on $Y$ but is strongly correlated in training; if the contrastive intervention loss still improves out-of-distribution performance compared to ERM, then the improvement cannot be attributed to backdoor adjustment, or the mixed representation assumption must be violated in a way that happens to help.

Watch

Extended reading notes

Core claim

The paper claims to be the first to explicitly treat confounding features as predictive in graph regression, and to show that doing so improves generalization. Its central proposal is a two-part loss: an enhanced GIB objective $-I(C;Y)+\alpha I(C;G)-\beta I(S;Y)$ that keeps the confounder $S$ informative for the target $Y$ while still compressing the causal subgraph $C$, and a contrastive intervention loss $\mathcal{L}_{\mathrm{CI}}$ that aligns the original graph representation with randomly mixed causal-and-confounding representations. The mixed representation $H_{\mathrm{mix},ij}=H_{c,i}+H_{s,j}$ is treated as a counterfactual graph, and the InfoNCE-style alignment is presented as an implicit backdoor adjustment in representation space. The paper reports that this recipe yields state-of-the-art out-of-distribution performance on graph regression benchmarks, improving over prior causal graph learning methods that discard confounding information.

Load-bearing premise

The method treats the vector sum of a causal representation and a confounding representation as a valid counterfactual graph, and assumes that aligning with such sums implements backdoor adjustment; if that mixed representation is not a faithful stand-in for intervening on the confounder, the claimed causal invariance does not follow.

Editorial extensions

If this is right

  • If the framework generalizes as reported, graph regression under distribution shift no longer needs to throw away confounders; modeling their predictive role should become a standard component of causal graph learning.
  • The contrastive intervention loss provides a label-free way to implement backdoor adjustment in representation space, which could make causal intervention applicable to continuous-label tasks beyond graphs.
  • The enhanced GIB objective suggests that information-theoretic disentanglement in regression should explicitly budget for confounder information rather than assuming it is noise.
  • The reported benchmark results, if reproduced, would give practitioners a default causal method for graph regression rather than forcing them to adapt classification-specific tools.

Reading between the lines

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

  • The random addition $H_{c,i}+H_{s,j}$ is a representation-space counterfactual; whether it faithfully reflects the underlying graph structure is not proven, so the method may be most reliable when causal and confounding substructures are separable in embedding space.
  • The contrastive alignment could be reinterpreted as a regularizer that encourages the causal representation to be invariant to confounder changes, which suggests a direct connection to invariant risk minimization that the paper does not explore.
  • A testable extension is to apply the same two-part loss to node-level or link-level regression, where the definition of a causal subgraph may need a different mask structure.
  • One could probe the claim about confounder predictiveness by measuring the correlation between the learned confounder representation and the target on a held-out set where the spurious correlation is intentionally broken.
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

2 major / 4 minor

Summary. This paper studies causal graph regression (CGR), i.e., graph-level regression under out-of-distribution shifts, and proposes a framework that (i) extends the graph information bottleneck objective by explicitly modeling the predictive power of confounding subgraphs through an additive term -βI(S;Y), and (ii) introduces a contrastive intervention loss that aligns an original graph representation with a representation formed by adding the causal embedding of one graph to the confounding embedding of another graph. The authors argue that this implements an implicit form of backdoor adjustment in representation space, and they report state-of-the-art OOD performance on the GOOD-ZINC benchmark and competitive results on ReactionOOD S-OOD datasets. The paper includes ablations and a parameter sensitivity analysis, and the code is publicly released.

Significance. If the central claims hold, the paper makes a useful contribution by adapting causal graph learning ideas to regression, a setting where label-based intervention techniques do not apply. The variational bounds for I(C;G) are standard, and the reduction of the mutual information terms to MSE under a constant-variance Gaussian assumption is internally consistent (apart from a presentation issue in Eqs. (13)-(14)). The empirical results are strong: the proposed method outperforms all baselines on all eight GOOD-ZINC settings and is best or second-best in most ReactionOOD settings. The ablations (Figure 4 and Appendix A.5) show that both the enhanced GIB loss and the contrastive intervention loss contribute to the gains. The code release is a definite strength. The main weakness is that the causal intervention mechanism, which is central to the paper's framing, rests on an unvalidated assumption that embedding addition produces a valid counterfactual representation; without a derivation or direct validation, the OOD improvement could be attributed to a generic contrastive regularizer.

major comments (2)
  1. [Section 4.3, Eq. (16)] The counterfactual representation Hmix,ij = Hc,i + Hs,j is asserted without justification. For a GNN encoder, the embedding of a graph with subgraphs C and S is f(A⊙M, X⊙M), which is not generally equal to f(C) + f(S). The paper states that this addition 'can be understood as an implicit realization of backdoor adjustment', but no derivation or empirical validation is provided. Since the causal intervention claim is load-bearing for the paper's title and contributions, the authors should either (a) provide a theoretical condition under which GNN embeddings are additive across subgraphs, or (b) run an ablation that isolates the additive counterfactual assumption, e.g., replacing Hmix with a non-additive mixing operation (such as concatenation or a learned combination) or with Hc,i plus random noise. Without such evidence, the OOD gains can be explained by a generic contrastive regularizer that encourages invariance to confounders, and the paper's causal graph regression interpretation is not established.
  2. [Section 5.2, Tables 1-3] The paper claims to be the first to explicitly consider the predictive role of confounding features in graph regression, but the experiments do not include any baseline that adapts existing causal graph learning methods (e.g., CAL or DisC) to regression. The introduction states that vanilla adaptation of CGL to regression is dwarfed by ERM, yet no such adaptation appears in the comparison tables. Without this baseline, it is unclear whether the observed improvements come from the proposed enhanced GIB objective, from the contrastive intervention, or simply from the architectural choices shared with the baselines. Adding an adapted CAL-style baseline, or at least an ablation that removes only the -βI(S;Y) term while retaining the rest of the framework, would strengthen the claim that modeling confounder predictiveness is the key novelty.
minor comments (4)
  1. [Section 4.2, Eq. (14)] The derivation of Eq. (14) is presented in a way that confuses mutual information with conditional entropy. Eq. (13) approximates H(Y|C), not I(C;Y). Since I(C;Y) = H(Y) - H(Y|C) and H(Y) is constant, maximizing I(C;Y) is equivalent to minimizing H(Y|C), which is approximately the squared error. The text should say 'the objective for maximizing I(C;Y) reduces to minimizing the least-squares loss' rather than 'I(C;Y) reduces to the least-squares loss' and should not write LCP as a negative MSE without clarifying the target is to minimize -LCP.
  2. [Section 4.3, Eq. (17)] The InfoNCE formula in Eq. (17) is missing a closing parenthesis in the denominator; the intended expression is -1/B Σ_i log [ exp(sim(Hg,i, Hmix,ij)) / Σ_{k≠i} exp(sim(Hg,i, Hg,k)) ]. The notation should be fixed for clarity.
  3. [Section 5.1, title] The subsection title contains a typo: 'Datsets' should be 'Datasets'.
  4. [Section 4.2, Eq. (5)-(8)] When deriving the upper bound for I(C;G), the notation µϕ(G) is used both for the mean of the conditional distribution p(C|G) and, in Eq. (8), as the quantity whose squared norm is averaged. Since the covariance is assumed to be the identity, this is acceptable, but the text could clarify that µϕ(G) is the mean vector of the embedding distribution.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the enhanced GIB losses are derived from variational bounds and the OOD claims are tested on external benchmarks, not forced by the paper's own definitions.

full rationale

The paper's derivation chain is self-contained and not circular. The enhanced GIB objective in Eq. (4) is decomposed into variational approximations: I(C;G) is bounded via a Gaussian assumption leading to the L2 proxy in Eq. (8), while I(C;Y) and I(S;Y) are reduced to least-squares losses through conditional Gaussian modeling in Eqs. (13)-(15). These are standard variational reductions rather than fitted parameters being renamed as predictions. The causal intervention loss in Eq. (17) is an InfoNCE objective that aligns original graph representations with mixed representations Hc,i + Hs,j from Eq. (16); the paper explicitly phrases the backdoor-adjustment connection as 'can be understood as an implicit realization' (Section 4.3), which is an interpretive analogy rather than a derivation whose conclusion is assumed in its premise. No parameter is fitted to the reported OOD targets and then called a prediction: the hyperparameters are fixed at 0.5 (Appendix A.4), and the method's generalization claims are evaluated on external benchmarks (GOOD-ZINC and ReactionOOD). The only self-citation is to the ReactionOOD benchmark paper, which is a dataset source and is not load-bearing for the method's causal derivation. There is no imported uniqueness theorem and no ansatz smuggled in through self-citation. The conceptual concern that LCI enforces the very invariance it claims to discover is a support/correctness question about the backdoor-adjustment interpretation, not a circularity of the empirical claims, which stand or fall on the external OOD results.

Assumptions & free parameters 3 free parameters · 6 assumptions · 1 invented entities

The central claim rests on several modeling assumptions: the Gaussian/identity-covariance approximation for the mutual information bounds, the constant-variance Gaussian model for the conditional response distributions, the SCM structure in Figure 1, and the heuristic that embedding addition plus InfoNCE implements causal intervention. The three loss weights (alpha, beta, lambda) are fixed by hand. The mixed representation Hmix is an invented training-only construct without independent evidence.

free parameters (3)
  • alpha = 0.5
    Weight of the I(C;G) compression term in LGIB (Eq. 4). Set to 0.5 for all experiments.
  • beta = 0.5
    Weight of the I(S;Y) term in LGIB (Eq. 4). Set to 0.5 for all experiments; sensitivity analysis suggests an optimal range around 0.3-0.6.
  • lambda = 0.5
    Weight of the causal intervention loss LCI (Remark 4.1). Set to 0.5 for all experiments; sensitivity analysis suggests an optimal range around 0.2-0.4.
assumptions (6)
  • domain assumption The posterior p(C|G) is Gaussian with identity covariance (Eq. 6-7).
    Used to compute the variational upper bound of I(C;G) as 1/2 E||mu||^2. The paper claims any full-rank covariance can be whitened, but the identity-covariance assumption is a modeling simplification.
  • domain assumption The conditional distributions p(Y|Hc) and p(Y|Hs) have constant variance sigma^2=1 (Eq. 11-14).
    This assumption reduces -H(Y|C) and -H(Y|S) to least-squares losses (Eq. 13-14), which is central to the proposed loss's form.
  • domain assumption The SCM in Figure 1: G generates C and S, C affects Y, and there is a backdoor path S -> C -> Y.
    This motivates the inclusion of -beta*I(S;Y) and the claim that S has predictive power for Y.
  • domain assumption Confounding subgraphs S have predictive power for Y in real graph regression tasks.
    The paper's core hypothesis; used to justify adding I(S;Y) to the objective. The paper supports it with a real-world example (molecular weight and toxicity) but does not provide a formal causal argument.
  • ad hoc to paper The random addition Hmix = Hc + Hs yields a valid counterfactual or intervened representation.
    Used in Eq. (16) to construct positive pairs for the contrastive loss. No theoretical or empirical validation that embedding addition corresponds to a graph-level intervention.
  • ad hoc to paper Optimizing the InfoNCE loss (Eq. 17) with these positive pairs enforces causal invariance.
    The paper states this is an 'implicit realization of backdoor adjustment' but does not derive the equivalence.
invented entities (1)
  • Hmix,ij (mixed graph representation)
    purpose: A positive-pair representation for the contrastive loss, created by adding a causal subgraph embedding Hc,i and a confounding subgraph embedding Hs,j. Used to train the model to be invariant to the confounding subgraph.
    This is a synthetic embedding constructed solely for training; it has no external observable counterpart and its validity as a counterfactual is assumed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Recipe for Causal Graph Regression: Confounding Effects Revisited." pith.science (2026). https://pith.science/paper/ZX7YSV2I

@misc{pith2026250700440,
  author       = {Pith},
  title        = {Pith review of: A Recipe for Causal Graph Regression: Confounding Effects Revisited},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZX7YSV2I}},
  note         = {Machine review of arXiv:2507.00440}
}
read the original abstract

Through recognizing causal subgraphs, causal graph learning (CGL) has risen to be a promising approach for improving the generalizability of graph neural networks under out-of-distribution (OOD) scenarios. However, the empirical successes of CGL techniques are mostly exemplified in classification settings, while regression tasks, a more challenging setting in graph learning, are overlooked. We thus devote this work to tackling causal graph regression (CGR); to this end we reshape the processing of confounding effects in existing CGL studies, which mainly deal with classification. Specifically, we reflect on the predictive power of confounders in graph-level regression, and generalize classification-specific causal intervention techniques to regression through a lens of contrastive learning. Extensive experiments on graph OOD benchmarks validate the efficacy of our proposals for CGR. The model implementation and the code are provided on https://github.com/causal-graph/CGR.

Figures

Figures reproduced from arXiv: 2507.00440 by the authors.

Figure 1
Figure 1. Structural causal model (SCM) for graph regression. The core methodology of causal learning involves the identi￾fication and differentiation of causal features from confound￾ing ones. As shown in [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Given a mini-batch of graphs, (1) the GNN encoder computes the graph embeddings Hg, and an attention layer generates soft masks to extract causal and confounding subgraphs. (2) GNN Gc processes the causal subgraph C, generates its representation Hc, and employs readout to predict responses; it is optimized with causal subgraph loss Lc. (3) GNN Gs, sharing parameters with Gc, processes the confounding subgraph S, gen… view at source ↗
Figure 3
Figure 3. Ablation study on confounder predictive power (left) and causal intervention methods (right) for OOD generalization on GOOD-Motif. 6. Conclusion In this work, we propose a recipe for causal graph regres￾sion through reshaping the processing of confounding ef￾fects in existing CGL classification-specific techniques. In particular, we develop an enhanced graph information bot￾tleneck (GIB) loss function which highligh… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The comparison of different components. evaluated using MAE, while the other datasets adopt RMSE as the evaluation metric. Given that the ZINC results are small (approximately 0.0x), we scale them by a factor of 10 in the [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: Parameter sensitivity. indicate that, there is no clear trend toward getting better or worse for α. For β, which balances the GIB loss, there is a gradual increase in RMSE when it is too large, especially in scaffold-covariate settings, suggesting an optimal range arou…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

54 extracted references · 39 canonical work pages

  1. [1]

    Deep evidential regression

    Amini, A., Schwarting, W., Soleimany, A., and Rus, D. Deep evidential regression. Advances in neural information processing systems, 33: 0 14927--14937, 2020

  2. [2]

    Invariant risk minimization

    Arjovsky, M., Bottou, L., Gulrajani, I., and Lopez-Paz, D. Invariant risk minimization. arXiv preprint arXiv:1907.02893, 2019

  3. [3]

    Information bottleneck for gaussian variables

    Chechik, G., Globerson, A., Tishby, N., and Weiss, Y. Information bottleneck for gaussian variables. Advances in Neural Information Processing Systems, 16, 2003

  4. [4]

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

    Chen, Y., Zhang, Y., Bian, Y., Yang, H., Kaili, M., Xie, B., Liu, T., Han, B., and Cheng, J. Learning causally invariant representations for out-of-distribution generalization on graphs. Advances in Neural Information Processing Systems, 35: 0 22131--22148, 2022

  5. [5]

    Does invariant graph learning via environment augmentation learn invariance? Advances in Neural Information Processing Systems, 36, 2024

    Chen, Y., Bian, Y., Zhou, K., Xie, B., Han, B., and Cheng, J. Does invariant graph learning via environment augmentation learn invariance? Advances in Neural Information Processing Systems, 36, 2024

  6. [6]

    Debiasing graph neural networks via learning disentangled causal substructure

    Fan, S., Wang, X., Mo, Y., Shi, C., and Tang, J. Debiasing graph neural networks via learning disentangled causal substructure. Advances in Neural Information Processing Systems, 35: 0 24934--24946, 2022

  7. [7]

    Generalizing graph neural networks on out-of-distribution graphs

    Fan, S., Wang, X., Shi, C., Cui, P., and Wang, B. Generalizing graph neural networks on out-of-distribution graphs. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

  8. [8]

    Domain-adversarial training of neural networks

    Ganin, Y., Ustinova, E., Ajakan, H., Germain, P., Larochelle, H., Laviolette, F., March, M., and Lempitsky, V. Domain-adversarial training of neural networks. Journal of machine learning research, 17 0 (59): 0 1--35, 2016

Show all 54 references
  1. [9]

    Robust causal graph representation learning against confounding effects

    Gao, H., Li, J., Qiang, W., Si, L., Xu, B., Zheng, C., and Sun, F. Robust causal graph representation learning against confounding effects. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, pp.\ 7624--7632, 2023

  2. [10]

    N., Duvenaud, D., Hern \'a ndez-Lobato, J

    G \'o mez-Bombarelli, R., Wei, J. N., Duvenaud, D., Hern \'a ndez-Lobato, J. M., S \'a nchez-Lengeling, B., Sheberla, D., Aguilera-Iparraguirre, J., Hirzel, T. D., Adams, R. P., and Aspuru-Guzik, A. Automatic chemical design using a data-driven continuous representation of mol...

  3. [11]

    Good: A graph out-of-distribution benchmark

    Gui, S., Li, X., Wang, L., and Ji, S. Good: A graph out-of-distribution benchmark. Advances in Neural Information Processing Systems, 35: 0 2059--2073, 2022

  4. [12]

    R., and Liu, H

    Guo, R., Cheng, L., Li, J., Hahn, P. R., and Liu, H. A survey of learning causality with data: Problems and methods. ACM Computing Surveys (CSUR), 53 0 (4): 0 1--37, 2020

  5. [13]

    Counterfactual learning on graphs: A survey

    Guo, Z., Wu, Z., Xiao, T., Aggarwal, C., Liu, H., and Wang, S. Counterfactual learning on graphs: A survey. Machine Intelligence Research, 22 0 (1): 0 17--59, 2025

  6. [14]

    Junction tree variational autoencoder for molecular graph generation

    Jin, W., Barzilay, R., and Jaakkola, T. Junction tree variational autoencoder for molecular graph generation. In International conference on machine learning, pp.\ 2323--2332. PMLR, 2018

  7. [15]

    P., Welling, M., et al

    Kingma, D. P., Welling, M., et al. Auto-encoding variational bayes, 2013

  8. [16]

    Out-of-distribution generalization via risk extrapolation (rex)

    Krueger, D., Caballero, E., Jacobsen, J.-H., Zhang, A., Binas, J., Zhang, D., Le Priol, R., and Courville, A. Out-of-distribution generalization via risk extrapolation (rex). In International conference on machine learning, pp.\ 5815--5826. PMLR, 2021

  9. [17]

    J., Paige, B., and Hern \'a ndez-Lobato, J

    Kusner, M. J., Paige, B., and Hern \'a ndez-Lobato, J. M. Grammar variational autoencoder. In International conference on machine learning, pp.\ 1945--1954. PMLR, 2017

  10. [18]

    L., and Van Lint, H

    Li, G., Knoop, V. L., and Van Lint, H. Multistep traffic forecasting by dynamic graph convolution: Interpretations of real-time spatial correlations. Transportation Research Part C: Emerging Technologies, 128: 0 103185, 2021

  11. [19]

    Ood-gnn: Out-of-distribution generalized graph neural network

    Li, H., Wang, X., Zhang, Z., and Zhu, W. Ood-gnn: Out-of-distribution generalized graph neural network. IEEE Transactions on Knowledge and Data Engineering, 35 0 (7): 0 7328--7340, 2022

  12. [20]

    Generative causal explanations for graph neural networks

    Lin, W., Lan, H., and Li, B. Generative causal explanations for graph neural networks. In International Conference on Machine Learning, pp.\ 6666--6679. PMLR, 2021

  13. [21]

    Parameterized explainer for graph neural network

    Luo, D., Cheng, W., Xu, D., Yu, W., Zong, B., Chen, H., and Zhang, X. Parameterized explainer for graph neural network. Advances in neural information processing systems, 33: 0 19620--19631, 2020

  14. [22]

    Utilizing reinforcement learning and causal graph networks to address the intricate dynamics in financial risk prediction

    Ma, F., Li, H., and Ilyas, M. Utilizing reinforcement learning and causal graph networks to address the intricate dynamics in financial risk prediction. International Journal of Information Technologies and Systems Approach (IJITSA), 17 0 (1): 0 1--19, 2024

  15. [23]

    A survey of out-of-distribution generalization for graph machine learning from a causal view

    Ma, J. A survey of out-of-distribution generalization for graph machine learning from a causal view. arXiv preprint arXiv:2409.09858, 2024

  16. [24]

    Interpretable and generalizable graph learning via stochastic attention mechanism

    Miao, S., Liu, M., and Li, P. Interpretable and generalizable graph learning via stochastic attention mechanism. In International Conference on Machine Learning, pp.\ 15524--15543. PMLR, 2022

  17. [25]

    Representation learning via invariant causal mechanisms

    Mitrovic, J., McWilliams, B., Walker, J., Buesing, L., and Blundell, C. Representation learning via invariant causal mechanisms. arXiv preprint arXiv:2010.07922, 2020

  18. [26]

    and Weigend, A

    Nix, D. and Weigend, A. Learning local error bars for nonlinear regression. Advances in neural information processing systems, 7, 1994

  19. [27]

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

  20. [28]

    Interpretation and identification of causal mediation

    Pearl, J. Interpretation and identification of causal mediation. Psychological methods, 19 0 (4): 0 459, 2014

  21. [29]

    Pleiss, G., Souza, A., Kim, J., Li, B., and Weinberger, K. Q. Neural network out-of-distribution detection for regression tasks. 2019

  22. [30]

    Causal enhanced drug--target interaction prediction based on graph generation and multi-source information fusion

    Qiao, G., Wang, G., and Li, Y. Causal enhanced drug--target interaction prediction based on graph generation and multi-source information fusion. Bioinformatics, 40 0 (10): 0 btae570, 2024

  23. [31]

    A., Cheng, A

    Rollins, Z. A., Cheng, A. C., and Metwally, E. Molprop: Molecular property prediction with multimodal language and graph fusion. Journal of Cheminformatics, 16 0 (1): 0 56, 2024

  24. [32]

    The perceptron: a probabilistic model for information storage and organization in the brain

    Rosenblatt, F. The perceptron: a probabilistic model for information storage and organization in the brain. Psychological review, 65 0 (6): 0 386, 1958

  25. [33]

    W., Hashimoto, T

    Sagawa, S., Koh, P. W., Hashimoto, T. B., and Liang, P. Distributionally robust neural networks for group shifts: On the importance of regularization for worst-case generalization. arXiv preprint arXiv:1911.08731, 2019

  26. [34]

    Spiekermann, K., Pattanaik, L., and Green, W. H. High accuracy barrier heights, enthalpies, and rate coefficients for chemical reactions. Scientific Data, 9 0 (1): 0 417, 2022

  27. [35]

    Stuyver, T., Jorner, K., and Coley, C. W. Reaction profiles for quantum chemistry-computed [3+ 2] cycloaddition reactions. Scientific Data, 10 0 (1): 0 66, 2023

  28. [36]

    Causal attention for interpretable and generalizable graph classification

    Sui, Y., Wang, X., Wu, J., Lin, M., He, X., and Chua, T.-S. Causal attention for interpretable and generalizable graph classification. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pp.\ 1696--1705, 2022

  29. [37]

    Enhancing out-of-distribution generalization on graphs via causal attention learning

    Sui, Y., Mao, W., Wang, S., Wang, X., Wu, J., He, X., and Chua, T.-S. Enhancing out-of-distribution generalization on graphs via causal attention learning. ACM Transactions on Knowledge Discovery from Data, 18 0 (5): 0 1--24, 2024

  30. [38]

    and Saenko, K

    Sun, B. and Saenko, K. Deep coral: Correlation alignment for deep domain adaptation. In Computer Vision--ECCV 2016 Workshops: Amsterdam, The Netherlands, October 8-10 and 15-16, 2016, Proceedings, Part III 14, pp.\ 443--450. Springer, 2016

  31. [39]

    M., and Liang, P

    Tajwar, F., Kumar, A., Xie, S. M., and Liang, P. No true state-of-the-art? ood detection methods are inconsistent across datasets. arXiv preprint arXiv:2109.05554, 2021

  32. [40]

    and Zaslavsky, N

    Tishby, N. and Zaslavsky, N. Deep learning and the information bottleneck principle. In 2015 ieee information theory workshop (itw), pp.\ 1--5. IEEE, 2015

  33. [41]

    C., and Bialek, W

    Tishby, N., Pereira, F. C., and Bialek, W. The information bottleneck method. arXiv preprint physics/0004057, 2000

  34. [42]

    Principles of risk minimization for learning theory

    Vapnik, V. Principles of risk minimization for learning theory. Advances in neural information processing systems, 4, 1991

  35. [43]

    F., Heinen, S

    von Rudorff, G. F., Heinen, S. N., Bragato, M., and von Lilienfeld, O. A. Thousands of reactants and transition states for competing e2 and s2 reactions. Machine Learning: Science and Technology, 1 0 (4): 0 045026, 2020

  36. [44]

    and Veitch, V

    Wang, Z. and Veitch, V. A unified causal view of domain invariant representation learning. 2022

  37. [45]

    Towards out-of-distribution generalizable predictions of chemical kinetics properties

    Wang, Z., Chen, Y., Duan, Y., Li, W., Han, B., Cheng, J., and Tong, H. Towards out-of-distribution generalizable predictions of chemical kinetics properties. arXiv preprint arXiv:2310.03152, 2023

  38. [46]

    Handling distribution shifts on graphs: An invariance perspective

    Wu, Q., Zhang, H., Yan, J., and Wipf, D. Handling distribution shifts on graphs: An invariance perspective. arXiv preprint arXiv:2202.02466, 2022 a

  39. [47]

    Graph information bottleneck

    Wu, T., Ren, H., Li, P., and Leskovec, J. Graph information bottleneck. Advances in Neural Information Processing Systems, 33: 0 20437--20448, 2020

  40. [48]

    Discovering invariant rationales for graph neural networks

    Wu, Y.-X., Wang, X., Zhang, A., He, X., and Chua, T.-S. Discovering invariant rationales for graph neural networks. arXiv preprint arXiv:2201.12872, 2022 b

  41. [49]

    Deconfounding to explanation evaluation in graph neural networks

    Wu, Y.-X., Wang, X., Zhang, A., Hu, X., Feng, F., He, X., and Chua, T.-S. Deconfounding to explanation evaluation in graph neural networks. arXiv preprint arXiv:2201.08802, 2022 c

  42. [50]

    Yu, S., Yu, X., L kse, S., Jenssen, R., and Pr \' ncipe, J. C. Cauchy-schwarz divergence information bottleneck for regression. In ICLR, 2024

  43. [51]

    mixup: Beyond empirical risk minimization

    Zhang, H. mixup: Beyond empirical risk minimization. arXiv preprint arXiv:1710.09412, 2017

  44. [52]

    Regexplainer: Generating explanations for graph neural networks in regression task

    Zhang, J., Chen, Z., Mei, H., Luo, D., and Wei, H. Regexplainer: Generating explanations for graph neural networks in regression task. arXiv preprint arXiv:2307.07840, 2023

  45. [53]

    Zhao, S., Prapas, I., Karasante, I., Xiong, Z., Papoutsis, I., Camps-Valls, G., and Zhu, X. X. Causal graph neural networks for wildfire danger prediction. arXiv preprint arXiv:2403.08414, 2024

  46. [54]

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