• Corpus ID: 235755509

Nested Counterfactual Identification from Arbitrary Surrogate Experiments

@inproceedings{Correa2021NestedCI,
  title={Nested Counterfactual Identification from Arbitrary Surrogate Experiments},
  author={Juan David Correa and Sanghack Lee and Elias Bareinboim},
  booktitle={NeurIPS},
  year={2021}
}
The Ladder of Causation describes three qualitatively different types of activities an agent may be interested in engaging in, namely, seeing (observational), doing (interventional), and imagining (counterfactual) (Pearl and Mackenzie, 2018). The inferential challenge imposed by the causal hierarchy is that data is collected by an agent observing or intervening in a system (layers 1 and 2), while its goal may be to understand what would have happened had it taken a different course of action… 

Figures from this paper

Partial Counterfactual Identification from Observational and Experimental Data
TLDR
This paper shows that all counterfactual distributions in an arbitrary structural causal model (SCM) with discrete observed domains could be generated by a canonical family of SCMs with the same causal diagram where unobserved (exogenous) variables are also discrete, taking values in finite domains.
Counterfactual Transportability: A Formal Approach
TLDR
This paper investigates the transportability of counterfactuals from an arbitrary combination of observational and experimental distributions coming from disparate domains and introduces a sufficient and necessary graphical condition and develops an complete algorithm for transportingcounterfactual quantities across domains in nonparametric settings.
Causal Inference Through the Structural Causal Marginal Problem
TLDR
This work formalises an approach to counterfactual inference based on merging information from multiple datasets using the response function formulation and shows that it reduces the space of allowed marginal and joint SCMs.

References

SHOWING 1-10 OF 29 REFERENCES
Effects of Treatment on the Treated: Identification and Generalization
TLDR
The conditions under which ETT can be estimated from experimental and/or observational studies are explored, and it is shown that, when the action invokes a singleton variable, the conditions for ETT identification have simple characterizations in terms of causal diagrams.
On Pearl’s Hierarchy and the Foundations of Causal Inference
TLDR
This chapter develops a novel and comprehensive treatment of the Pearl Causal Hierarchy through two complementary lenses: one logical-probabilistic and another inferential-graphical, and investigates an inferential system known as do-calculus, showing how it can be suf­ ficient, and in many cases necessary, to allow inferences across the PCH’s layers.
General Identifiability with Arbitrary Surrogate Experiments
TLDR
A general strategy to prove non-gID based on hedgelets and thickets is introduced, which leads to a necessary and sufficient graphical condition for the corresponding decision problem and it is shown that do-calculus is complete for the task of g-identifiability.
Fairness in Decision-Making - The Causal Explanation Formula
TLDR
The causal explanation formula is derived, which allows the AI designer to quantitatively evaluate fairness and explain the total observed disparity of decisions through different discriminatory mechanisms, and provides a quantitative approach to policy implementation and the design of fair AI systems.
Causal Effect Identifiability under Partial-Observability
TLDR
This paper studies a number of properties of the factors that comprise a causal effect under various levels of abstraction, and establishes a sufficient graphical criterion for determining whether the effects are identifiable from partially-observed distributions.
Causal Inference by Surrogate Experiments: z-Identifiability
TLDR
The problem of estimating the effect of intervening on a set of variables X from experiments on a different set, Z, that is more accessible to manipulation is addressed and a graphical necessary and sufficient condition for z-identifiability for arbitrary sets X,Z, and Y is provided.
Identifiability in Causal Bayesian Networks: A Sound and Complete Algorithm
TLDR
It is shown that the identify algorithm that Tian and Pearl define and prove sound for semi-Markovian models can be transfered to general causal graphs and is not only sound, but also complete, effectively solving the identifiability question for causal Bayesian networks.
Identifiability of Path-Specific Effects
TLDR
Simple, graphical conditions for experimental identifiability of path-specific effects are derived, namely, conditions under which path- specific effects can be estimated consistently from data obtained from controlled experiments.
A general identification condition for causal effects
TLDR
The paper establishes a necessary and sufficient criterion for the identifiability of the causal effects of a singleton variable on all other variables in the model, and a powerful sufficient criterion on any set of variables.
Identification of Personalized Effects Associated With Causal Pathways
TLDR
This paper defines counterfactual responses to such policies, gives a general identification algorithm for thesecounterfactuals, and proves completeness of the algorithm for unrestricted policies.
...
...