The reviewed record of science sign in
Pith

arxiv: 2605.24358 · v4 · pith:RXDR62AN · submitted 2026-05-23 · cs.LG · cs.AI

Treatment Effect Estimation with Differentiated Networked Effect on Graph Data

Xiaofeng Lin , Han Bao , Hisashi Kashima This is my paper

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel 2026-06-30 14:59 UTCgrok-4.3pith:RXDR62ANrecord.jsonopen to challenge →

classification cs.LG cs.AI
keywords individual treatment effectgraph datanetworked interferencedifferentiated networked effectpartial attentionmessage passingobservational datacausal inference
0
0 comments X

The pith

Differentiated networked effect from neighbors of varying importance and scale must be modeled explicitly to avoid imprecise individual treatment effect estimates on graphs.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper claims that interference in observational graph data for individual treatment effect estimation includes a previously overlooked differentiated networked effect, in which neighbors contribute differently according to their learned importance and the size of their local networks. Existing interference models treat these contributions uniformly and therefore leave residual bias in the estimates. The authors introduce an interference modeling mechanism built from two partial attention layers that weight neighbor contributions and a message amplifier that rescales the aggregated signal according to neighborhood size. When these components are added, estimates on three real-world graphs improve over prior graph-based ITE methods. A reader should care because the bias from ignoring DNE can produce misguided policy or medical decisions that rely on networked observational data.

Core claim

The central claim is that differentiated networked effect is a critical component of interference on graphs, caused by local networks whose neighbors differ in importance and scale, and that an interference modeling mechanism using partial attention to estimate neighbor weights together with a message amplifier to adjust for scale produces more accurate ITE estimates than methods that do not differentiate these effects.

What carries the argument

The interference modeling mechanism built from two partial attention mechanisms that estimate neighbor importance and a message amplifier that adjusts the interference signal according to neighborhood scale.

If this is right

  • ITE estimates remain biased whenever neighbor contributions are aggregated without learned differentiation of importance or scale.
  • Decisions in commerce and medicine that rely on graph-structured observational data become more reliable once DNE is explicitly captured.
  • The same interference modeling components can be inserted into other graph neural architectures that estimate treatment effects under network interference.
  • Graphs whose local neighborhoods differ substantially in size will show larger accuracy gains from the message amplifier than more uniform graphs.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The partial attention approach could be adapted to dynamic or temporal graphs if the attention weights are allowed to evolve over time.
  • Similar differentiation of neighbor effects might improve other graph learning tasks such as influence maximization or contagion modeling where uniform aggregation is currently used.
  • Controlled experiments that systematically vary the degree of DNE in synthetic data would isolate how much of the reported gains come from the new components versus other modeling choices.

Load-bearing premise

Partial attention mechanisms can reliably learn the varying importance and scales of neighbors without introducing new biases or needing extra assumptions about how the data were generated.

What would settle it

On a synthetic graph dataset in which neighbor importance weights and local network sizes are known and controlled, the proposed model should recover the ground-truth ITE values more accurately than baselines that omit the partial attention and message amplifier components.

Figures

Figures reproduced from arXiv: 2605.24358 by Han Bao, Hisashi Kashima, Xiaofeng Lin.

Figure 1
Figure 1. Figure 1: Comparison of representations generated by im [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Architecture of GITE. Here, we show an example of three individuals. NIML represents an NIM layer. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: An example of a causal graph for an individual in [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 2
Figure 2. Figure 2 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Architecture of an NIM layer. We show an example where the underlying graph consists of three individuals. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Results (mean and standard errors) of sensitivity experiments for hyperparameters [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Results (mean and standard errors) of additional sensitivity experiments for hyperparameters [PITH_FULL_IMAGE:figures/full_fig_p022_6.png] view at source ↗
read the original abstract

Estimating individual treatment effect (ITE) from observational graph data is crucial for decision-making in the fields such as commerce and medicine. This task is challenging due to interference, where individual outcomes can be influenced by the treatments and covariates of their neighbors. Existing methods attempt to model such interference for accurate ITE estimation. However, a critical issue is often overlooked: differentiated networked effect (DNE), an effect caused by local networks consisting of neighbors with varying importance and scales. Capturing DNE is vital; otherwise, we will end up with imprecise ITE estimation due to an erroneous characterization of interference, which can result in misguided decisions. To address this challenge, we propose a novel interference modeling mechanism that incorporates two partial attention mechanisms and a message amplifier. The partial attention mechanisms automatically estimate the importance of different neighbors in contributing to interference, while the message amplifier adjusts the results of the interference modeling mechanism based on the scale of neighbors, all of which enables the model to capture DNE. Experiments on three real-world graphs demonstrate that our methods outperform existing approaches for ITE estimation from graph data, which corroborates the importance of explicitly capturing DNE.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper introduces the concept of differentiated networked effect (DNE) as an overlooked source of interference in individual treatment effect (ITE) estimation on graph data, where neighbors have varying importance and scales. It proposes a model using two partial attention mechanisms to estimate neighbor importance and a message amplifier to adjust for neighbor scales. Experiments on three real-world graphs are claimed to show that the proposed method outperforms existing approaches, supporting the need to explicitly capture DNE.

Significance. If the experimental claims hold after proper validation, the work could advance causal inference on graphs by providing a mechanism to handle heterogeneous interference effects, which is relevant for applications like medicine and commerce where network structure influences outcomes. The introduction of partial attention and message amplification as tools for DNE is a targeted contribution, but its impact depends on isolating these components from general model capacity.

major comments (3)
  1. [Experiments section (implied by abstract)] The central experimental claim (outperformance on three graphs corroborating DNE importance) is load-bearing but unsupported in the provided abstract and lacks any mention of component ablations, capacity-matched baselines, or sensitivity checks on the attention formulation. Without these, gains cannot be attributed specifically to DNE capture rather than increased model flexibility.
  2. [Introduction / Model section] The definition and formalization of DNE (varying neighbor importance and scales causing imprecise ITE) is introduced as a critical overlooked factor but appears ad-hoc without a precise mathematical characterization or derivation showing how standard GNN aggregations fail to capture it by construction.
  3. [Experiments] No details are supplied on the three real-world graphs, baselines, metrics (e.g., PEHE, ATE error), error bars, or statistical tests, making it impossible to assess whether the outperformance is robust or reproducible.
minor comments (2)
  1. [Model] Notation for the partial attention mechanisms and message amplifier should be defined with explicit equations early in the model section to improve clarity.
  2. [Abstract] The abstract would benefit from a brief statement of the specific metrics and baseline methods used in the experiments.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments, which highlight areas where the manuscript can be strengthened. We address each major comment point by point below, indicating where revisions will be made to improve clarity, rigor, and reproducibility while preserving the core contribution on differentiated networked effects.

read point-by-point responses

  1. Referee: The central experimental claim (outperformance on three graphs corroborating DNE importance) is load-bearing but unsupported in the provided abstract and lacks any mention of component ablations, capacity-matched baselines, or sensitivity checks on the attention formulation. Without these, gains cannot be attributed specifically to DNE capture rather than increased model flexibility.

    Authors: We agree that the abstract alone does not convey the experimental details needed to isolate the contribution of DNE modeling. The full manuscript reports results across three real-world graphs showing outperformance over baselines. To directly address attribution, we will add component ablations isolating the partial attention mechanisms and message amplifier, include capacity-matched baseline variants, and provide sensitivity checks on the attention formulation in the revised experiments section. revision: yes

  2. Referee: The definition and formalization of DNE (varying neighbor importance and scales causing imprecise ITE) is introduced as a critical overlooked factor but appears ad-hoc without a precise mathematical characterization or derivation showing how standard GNN aggregations fail to capture it by construction.

    Authors: The manuscript introduces DNE in the introduction and formalizes neighbor importance and scale within the proposed interference modeling mechanism in the model section. However, we acknowledge that an explicit derivation contrasting standard GNN aggregations (e.g., mean or attention-based) with DNE-induced bias is not fully elaborated. We will add this mathematical characterization and derivation in the revised introduction and model sections to clarify why existing approaches fail by construction. revision: partial

  3. Referee: No details are supplied on the three real-world graphs, baselines, metrics (e.g., PEHE, ATE error), error bars, or statistical tests, making it impossible to assess whether the outperformance is robust or reproducible.

    Authors: We agree these experimental details are necessary for assessing robustness. The full manuscript describes the three graphs, selected baselines, and primary metrics including PEHE. In the revision we will expand the experiments section with full dataset statistics, explicit metric definitions (PEHE and ATE error), error bars across runs, and statistical significance tests to ensure reproducibility. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical validation of proposed mechanism is self-contained

full rationale

The paper proposes partial attention mechanisms plus a message amplifier to capture differentiated networked effect (DNE) on graphs and reports outperformance on three real-world datasets. No equations, fitted parameters renamed as predictions, self-definitional constructions, or load-bearing self-citations appear in the abstract or described claims. The central assertion (explicit DNE modeling improves ITE estimation) rests on external experimental comparison rather than reducing to the model's own definitions or prior author work by construction. This is the normal case of an empirical methods paper whose derivation chain does not collapse internally.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The paper relies on standard causal inference assumptions for observational graph data and introduces DNE as a new modeling target. No explicit free parameters are named in the abstract.

axioms (1)
  • domain assumption Observational graph data contains sufficient information to estimate individual treatment effects once interference is properly modeled
    Core premise of all ITE estimation from graphs stated in the abstract.
invented entities (1)
  • Differentiated Networked Effect (DNE) no independent evidence
    purpose: To represent varying importance and scales of neighbors contributing to interference
    Introduced in the abstract as the critical overlooked factor causing imprecise ITE estimates.

pith-pipeline@v0.9.1-grok · 5731 in / 1261 out tokens · 44918 ms · 2026-06-30T14:59:23.017125+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

97 extracted references · 97 canonical work pages · 3 internal anchors

  1. [1]

    Shishir Adhikari and Elena Zheleva. 2025. Inferring individual direct causal effects under heterogeneous peer influence.Machine Learning114, 4 (2025), 113

    work page 2025

  2. [2]

    Aronow and Cyrus Samii

    Peter M. Aronow and Cyrus Samii. 2017. Estimating average causal effects under general interference, with application to a social network experiment.The Annals of Applied Statistics11 (2017), 1912–1947

    work page 2017

  3. [3]

    Usaid Awan, Marco Morucci, Vittorio Orlandi, Sudeepa Roy, Cynthia Rudin, and Alexander Volfovsky. 2020. Almost-matching-exactly for treatment effect estimation under network interference. InProceedings of the 23rd International Conference on Artificial Intelligence and Statistics. 3252–3262

    work page 2020

  4. [4]

    Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hinton. 2016. Layer normaliza- tion.arXiv preprint arXiv:1607.06450(2016)

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  5. [5]

    Guillaume Basse and Avi Feller. 2018. Analyzing two-stage experiments in the presence of interference.J. Amer. Statist. Assoc.113, 521 (2018), 41–55

    work page 2018

  6. [6]

    Ruichu Cai, Zeqin Yang, Weilin Chen, Yuguang Yan, and Zhifeng Hao. 2023. Generalization bound for estimating causal effects from observational network data. InProceedings of the 32nd ACM International Conference on Information and Knowledge Management. 163–172

    work page 2023

  7. [7]

    Serina Chang, Damir Vrabac, Jure Leskovec, and Johan Ugander. 2023. Estimat- ing geographic spillover effects of COVID-19 policies from large-scale mobility networks. InProceedings of the 37th AAAI Conference on Artificial Intelligence, Vol. 37. 14161–14169

    work page 2023

  8. [8]

    Weilin Chen, Ruichu Cai, Zeqin Yang, Jie Qiao, Yuguang Yan, Zijian Li, and Zhifeng Hao. 2024. Doubly Robust Causal Effect Estimation under Networked Interference via Targeted Learning. InProceedings of the 41st International Con- ference on Machine Learning

    work page 2024

  9. [9]

    James Cheng, Zechao Shang, Hong Cheng, Haixun Wang, and Jeffrey Xu Yu

    work page

  10. [10]

    K-reach: Who is in your small world.arXiv preprint arXiv:1208.0090(2012)

    work page internal anchor Pith review Pith/arXiv arXiv 2012

  11. [11]

    Zhixuan Chu, Stephen L Rathbun, and Sheng Li. 2021. Graph infomax adversarial learning for treatment effect estimation with networked observational data. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 176–184

    work page 2021

  12. [12]

    Gabriele Corso, Luca Cavalleri, Dominique Beaini, Pietro Liò, and Petar Veličković

    work page

  13. [13]

    Principal neighbourhood aggregation for graph nets.Advances in Neural Information Processing Systems33 (2020), 13260–13271

    work page 2020

  14. [14]

    Nicolas Courty, Rémi Flamary, Amaury Habrard, and Alain Rakotomamonjy

    work page

  15. [15]

    Joint distribution optimal transportation for domain adaptation.Advances in Neural Information Processing Systems30 (2017)

    work page 2017

  16. [16]

    Ziqiang Cui, Xing Tang, Yang Qiao, Bowei He, Liang Chen, Xiuqiang He, and Chen Ma. 2024. Treatment-Aware Hyperbolic Representation Learning for Causal Effect Estimation with Social Networks. InProceedings of the 2024 SIAM International Conference on Data Mining. 289–297

    work page 2024

  17. [17]

    Marco Cuturi. 2013. Sinkhorn distances: Lightspeed computation of optimal transport.Advances in Neural Information Processing Systems26 (2013)

    work page 2013

  18. [18]

    Airoldi, and Fabrizia Mealli

    Laura Forastiere, Edoardo M. Airoldi, and Fabrizia Mealli. 2021. Identification and Estimation of Treatment and Interference Effects in Observational Studies on Networks.J. Amer. Statist. Assoc.116, 534 (2021), 901–918

    work page 2021

  19. [19]

    Laura Forastiere, Fabrizia Mealli, Albert Wu, and Edoardo M Airoldi. 2022. Es- timating causal effects under network interference with Bayesian generalized propensity scores.Journal of Machine Learning Research23, 289 (2022), 1–61

    work page 2022

  20. [20]

    Dennis Frauen and Stefan Feuerriegel. 2022. Estimating individual treatment effects under unobserved confounding using binary instruments.arXiv preprint arXiv:2208.08544(2022)

    work page arXiv 2022

  21. [21]

    Dennis Frauen, Konstantin Hess, and Stefan Feuerriegel. 2024. Model-agnostic meta-learners for estimating heterogeneous treatment effects over time.arXiv preprint arXiv:2407.05287(2024)

    work page arXiv 2024

  22. [22]

    Arthur Gretton, Olivier Bousquet, Alex Smola, and Bernhard Schölkopf. 2005. Measuring Statistical Dependence with Hilbert-Schmidt norms. InProceedings of the 16th International Conference on Algorithmic Learning Theory. 63–77

    work page 2005

  23. [23]

    Ruocheng Guo, Jundong Li, Yichuan Li, K Selçuk Candan, Adrienne Raglin, and Huan Liu. 2021. Ignite: A minimax game toward learning individual treatment effects from networked observational data. InProceedings of the 29th International Conference on International Joint Conferences on Artificial Intelligence. 4534–4540

    work page 2021

  24. [24]

    Ruocheng Guo, Jundong Li, and Huan Liu. 2020. Learning individual causal effects from networked observational data. InProceedings of the 13th International Conference on Web Search and Data Mining. 232–240

    work page 2020

  25. [25]

    Xingzhuo Guo, Yuchen Zhang, Jianmin Wang, and Mingsheng Long. 2023. Esti- mating heterogeneous treatment effects: Mutual information bounds and learning algorithms. InProceedings of the 40th International Conference on Machine Learn- ing. 12108–12121

    work page 2023

  26. [26]

    Hamilton, Rex Ying, and Jure Leskovec

    William L. Hamilton, Rex Ying, and Jure Leskovec. 2017. Inductive Representation Learning on Large Graphs.Advances in Neural Information Processing Systems30 (2017)

    work page 2017

  27. [27]

    Shonosuke Harada and Hisashi Kashima. 2021. Graphite: Estimating individual ef- fects of graph-structured treatments. InProceedings of the 30th ACM International Conference on Information & Knowledge Management. 659–668

    work page 2021

  28. [28]

    Bowei He, Yunpeng Weng, Xing Tang, Ziqiang Cui, Zexu Sun, Liang Chen, Xi- uqiang He, and Chen Ma. 2024. Rankability-enhanced revenue uplift modeling framework for online marketing. InProceedings of the 30th ACM SIGKDD Confer- ence on Knowledge Discovery and Data Mining. 5093–5104

    work page 2024

  29. [29]

    Ruining He and Julian McAuley. 2016. Ups and downs: Modeling the visual evolution of fashion trends with one-class collaborative filtering. InProceedings of the 2016 World Wide Web Conference. 507–517

    work page 2016

  30. [30]

    Qiang Huang, Jing Ma, Jundong Li, Ruocheng Guo, Huiyan Sun, and Yi Chang

    work page

  31. [31]

    Modeling Interference for Individual Treatment Effect Estimation from Networked Observational Data.ACM Transactions on Knowledge Discovery from Data18, 3 (2023), 1–21

    work page 2023

  32. [32]

    Elizabeth Halloran

    Michael G Hudgens and M. Elizabeth Halloran. 2008. Toward Causal Inference With Interference.J. Amer. Statist. Assoc.103, 482 (2008), 832–842

    work page 2008

  33. [33]

    Andrew Jesson, Sören Mindermann, Yarin Gal, and Uri Shalit. 2021. Quantifying ignorance in individual-level causal-effect estimates under hidden confounding. InProceedings of the 38th International Conference on Machine Learning. 4829– 4838

    work page 2021

  34. [34]

    Andrew Jesson, Panagiotis Tigas, Joost van Amersfoort, Andreas Kirsch, Uri Shalit, and Yarin Gal. 2021. Causal-bald: Deep bayesian active learning of out- comes to infer treatment-effects from observational data.Advances in Neural Information Processing Systems34 (2021), 30465–30478

    work page 2021

  35. [35]

    Song Jiang, Zijie Huang, Xiao Luo, and Yizhou Sun. 2023. CF-GODE: Continuous- time causal inference for multi-agent dynamical systems. InProceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 997– 1009

    work page 2023

  36. [36]

    Song Jiang and Yizhou Sun. 2022. Estimating causal effects on networked ob- servational data via representation learning. InProceedings of the 31st ACM International Conference on Information & Knowledge Management. 852–861

    work page 2022

  37. [37]

    Fredrik Johansson, Uri Shalit, and David Sontag. 2016. Learning Representations for Counterfactual Inference. InProceedings of the 33rd International Conference on Machine Learning, Vol. 48. 3020–3029

    work page 2016

  38. [38]

    Fredrik D Johansson, Uri Shalit, Nathan Kallus, and David Sontag. 2022. General- ization bounds and representation learning for estimation of potential outcomes and causal effects.Journal of Machine Learning Research23, 166 (2022), 1–50

    work page 2022

  39. [39]

    Jean Kaddour, Yuchen Zhu, Qi Liu, Matt J Kusner, and Ricardo Silva. 2021. Causal effect inference for structured treatments.Advances in Neural Information Pro- cessing Systems34 (2021), 24841–24854

    work page 2021

  40. [40]

    Leonid V Kantorovich. 2006. On the Translocation of Masses.Journal of Mathe- matical Sciences133, 4 (2006)

    work page 2006

  41. [41]

    Diederik P Kingma and Jimmy Ba. 2015. Adam: A method for stochastic optimiza- tion. InProceedings of the 3rd International Conference on Learning Representations

    work page 2015

  42. [42]

    Kun Kuang, Lian Li, Zhi Geng, Lei Xu, Kun Zhang, Beishui Liao, Huaxin Huang, Peng Ding, Wang Miao, and Zhichao Jiang. 2020. Causal inference.Engineering 6, 3 (2020), 253–263

    work page 2020

  43. [43]

    Quoc Le and Tomas Mikolov. 2014. Distributed representations of sentences and documents. InProceedings of the 31st International Conference on Machine Learning. 1188–1196

    work page 2014

  44. [44]

    Haoxuan Li, Yan Lyu, Chunyuan Zheng, and Peng Wu. 2023. TDR-CL: Tar- geted Doubly Robust Collaborative Learning for Debiased Recommendations. In Proceedings of the 11th International Conference on Learning Representations

    work page 2023

  45. [45]

    Haoxuan Li, Chunyuan Zheng, and Peng Wu. 2023. StableDR: Stabilized Dou- bly Robust Learning for Recommendation on Data Missing Not at Random. In Proceedings of the 11th International Conference on Learning Representations

    work page 2023

  46. [46]

    Jundong Li, Ruocheng Guo, Chenghao Liu, and Huan Liu. 2019. Adaptive un- supervised feature selection on attributed networks. InProceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 92–100

    work page 2019

  47. [47]

    Jundong Li, Xia Hu, Jiliang Tang, and Huan Liu. 2015. Unsupervised streaming feature selection in social media. InProceedings of the 24th ACM International on Conference on Information and Knowledge Management. 1041–1050

    work page 2015

  48. [48]

    Xiaofeng Lin, Han Bao, Yan Cui, Koh Takeuchi, and Hisashi Kashima. 2025. Scalable individual treatment effect estimator for large graphs.Machine Learning 114, 1 (2025), 1–19. Treatment Effect Estimation with Differentiated Networked Effect on Graph Data KDD 2026, August 9–13, 2026, Jeju Island, Republic of Korea

    work page 2025

  49. [49]

    Xiaofeng Lin, Guoxi Zhang, Xiaotian Lu, Han Bao, Koh Takeuchi, and Hisashi Kashima. 2023. Estimating Treatment Effects Under Heterogeneous Interference. InJoint European Conference on Machine Learning and Knowledge Discovery in Databases. 576–592

    work page 2023

  50. [50]

    Chuang Liu, Yibing Zhan, Jia Wu, Chang Li, Bo Du, Wenbin Hu, Tongliang Liu, and Dacheng Tao. 2022. Graph pooling for graph neural networks: Progress, challenges, and opportunities.arXiv preprint arXiv:2204.07321(2022)

    work page arXiv 2022

  51. [51]

    Lan Liu and Michael G Hudgens. 2014. Large sample randomization inference of causal effects in the presence of interference.J. Amer. Statist. Assoc.109, 505 (2014), 288–301

    work page 2014

  52. [52]

    Meng Liu, Haiyang Yu, and Shuiwang Ji. 2024. Empowering GNNs via Edge- Aware Weisfeiler-Leman Algorithm.Transactions on Machine Learning Research (2024)

    work page 2024

  53. [53]

    Jing Ma, Yushun Dong, Zheng Huang, Daniel Mietchen, and Jundong Li. 2022. Assessing the causal impact of COVID-19 related policies on outbreak dynamics: A case study in the US. InProceedings of the 2022 Web Conference. 2678–2686

    work page 2022

  54. [54]

    Jing Ma, Ruocheng Guo, Chen Chen, Aidong Zhang, and Jundong Li. 2021. De- confounding with networked observational data in a dynamic environment. In Proceedings of the 14th ACM International Conference on Web Search and Data Mining. 166–174

    work page 2021

  55. [55]

    Jing Ma, Mengting Wan, Longqi Yang, Jundong Li, Brent Hecht, and Jaime Teevan

    work page

  56. [56]

    InProceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining

    Learning Causal Effects on Hypergraphs. InProceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 1202–1212

    work page

  57. [57]

    Liheng Ma, Chen Lin, Derek Lim, Adriana Romero-Soriano, Puneet K Dokania, Mark Coates, Philip Torr, and Ser-Nam Lim. 2023. Graph inductive biases in transformers without message passing. InProceedings of the 40th International Conference on Machine Learning. 23321–23337

    work page 2023

  58. [58]

    Yunpu Ma and Volker Tresp. 2021. Causal Inference under Networked Interfer- ence and Intervention Policy Enhancement. InProceedings of the 24th International Conference on Artificial Intelligence and Statistics, Vol. 130. 3700–3708

    work page 2021

  59. [59]

    Andrew L Maas, Awni Y Hannun, Andrew Y Ng, et al . 2013. Rectifier non- linearities improve neural network acoustic models. InProceedings of the 30th International Conference on Machine Learning, Vol. 30. 3

    work page 2013

  60. [60]

    Valentyn Melnychuk, Dennis Frauen, and Stefan Feuerriegel. 2023. Bounds on representation-induced confounding bias for treatment effect estimation.arXiv preprint arXiv:2311.11321(2023)

    work page arXiv 2023

  61. [61]

    Razieh Nabi, Joel Pfeiffer, Denis Charles, and Emre Kıcıman. 2022. Causal infer- ence in the presence of interference in sponsored search advertising.Frontiers in Big Data5 (2022)

    work page 2022

  62. [62]

    Miruna Oprescu, Jacob Dorn, Marah Ghoummaid, Andrew Jesson, Nathan Kallus, and Uri Shalit. 2023. B-learner: Quasi-oracle bounds on heterogeneous causal effects under hidden confounding. InProceedings of the 40th International Confer- ence on Machine Learning. 26599–26618

    work page 2023

  63. [63]

    Tian Qin, Tian-Zuo Wang, and Zhi-Hua Zhou. 2021. Budgeted heterogeneous treatment effect estimation. InProceedings of the 38th International Conference on Machine Learning. 8693–8702

    work page 2021

  64. [64]

    Vineeth Rakesh, Ruocheng Guo, Raha Moraffah, Nitin Agarwal, and Huan Liu

    work page

  65. [65]

    InProceedings of the 27th ACM International Conference on Information and Knowledge Management

    Linked causal variational autoencoder for inferring paired spillover effects. InProceedings of the 27th ACM International Conference on Information and Knowledge Management. 1679–1682

    work page

  66. [66]

    Paul R Rosenbaum. 2007. Interference between units in randomized experiments. J. Amer. Statist. Assoc.102, 477 (2007), 191–200

    work page 2007

  67. [67]

    Donald B Rubin. 1980. Randomization analysis of experimental data: The Fisher randomization test comment.J. Amer. Statist. Assoc.75, 371 (1980), 591–593

    work page 1980

  68. [68]

    Donald B Rubin. 2005. Causal inference using potential outcomes: Design, mod- eling, decisions.J. Amer. Statist. Assoc.100, 469 (2005), 322–331

    work page 2005

  69. [69]

    Mireille E Schnitzer. 2022. Estimands and Estimation of COVID-19 Vaccine Effectiveness Under the Test-Negative Design: Connections to Causal Inference. Epidemiology33, 3 (2022), 325

    work page 2022

  70. [70]

    Johansson, and David Sontag

    Uri Shalit, Fredrik D. Johansson, and David Sontag. 2017. Estimating individual treatment effect: generalization bounds and algorithms. InProceedings of the 34th International Conference on Machine Learning, Vol. 70. 3076–3085

    work page 2017

  71. [71]

    Claudia Shi, David Blei, and Victor Veitch. 2019. Adapting neural networks for the estimation of treatment effects.Advances in Neural Information Processing Systems32 (2019)

    work page 2019

  72. [72]

    Yongduo Sui, Caizhi Tang, Zhixuan Chu, Junfeng Fang, Yuan Gao, Qing Cui, Longfei Li, Jun Zhou, and Xiang Wang. 2024. Invariant Graph Learning for Treatment Effect Estimation from Networked Observational Data. InProceedings of the 2024 Web Conference

    work page 2024

  73. [73]

    Eric J Tchetgen Tchetgen and Tyler J VanderWeele. 2012. On causal inference in the presence of interference.Statistical Methods in Medical Research21, 1 (2012), 55–75

    work page 2012

  74. [74]

    Abhinav Thorat, Ravi Kolla, Niranjan Pedanekar, and Naoyuki Onoe. 2023. Es- timation of individual causal effects in network setup for multiple treatments. arXiv preprint arXiv:2312.11573(2023)

    work page arXiv 2023

  75. [75]

    Panos Toulis and Edward Kao. 2013. Estimation of causal peer influence effects. In Proceedings of the 30th International Conference on Machine Learning. 1489–1497

    work page 2013

  76. [76]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Ł ukasz Kaiser, and Illia Polosukhin. 2017. Attention is All You Need.Advances in Neural Information Processing Systems30 (2017)

    work page 2017

  77. [77]

    Victor Veitch, Yixin Wang, and David Blei. 2019. Using embeddings to correct for unobserved confounding in networks.Advances in Neural Information Processing Systems32 (2019)

    work page 2019

  78. [78]

    Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. 2017. Graph attention networks.arXiv preprint arXiv:1710.10903(2017)

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  79. [79]

    2008.Optimal transport: old and new

    Cédric Villani et al. 2008.Optimal transport: old and new. Vol. 338. Springer

    work page 2008

  80. [80]

    Davide Viviano. 2019. Policy Targeting under Network Interference.arXiv preprint arXiv:1906.10258(2019)

    work page arXiv 2019

Showing first 80 references.