← Back to all news
arXiv AI By Cl\'ement Yvernes, Marianne Clausel, Emilie Devijver, Eric Gaussier

Causal Effect Identifiability in Causal Abstractions: A Strict Hierarchy of Criteria

Read the original on arXiv AI →

The paper investigates how to determine causal effects when only partial causal information is available, using collections of compatible causal graphs instead of a single fully specified graph. It introduces several formal notions of identifiability for these graph collections and arranges them into a strict hierarchy that clarifies their interrelations. The authors show that a causal effect can be identified across a set of graphs even when no single causal derivation applies uniformly to all of them, thereby exposing gaps and ambiguities in prior work.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

Related stories

arXiv Machine Learning
Jun 19

Unsupervised Causal Abstractions Discovery

arXiv:2606. 19594v1 Announce Type: new Abstract: Causal abstractions formalize when a high-level structural causal model (SCM) captures the interventional behavior of a lower-level SCM.

By Th\'eo Saulus, Simon Lacoste-Julien, Dhanya Sridhar
More like this →
arXiv Machine Learning
Oct 5

Invariance of Clustering Operations in Causal Effect Identification

The paper studies how clustering variables in causal graphs can simplify inference but may also distort causal relations. It defines an "identification invariant" clustering operation—one that preserves both identifiability and nonidentifiability of causal effects—and presents a broad class of such operations based on conditions involving the c‑components of the original graph. The authors illustrate the practical application of these results in real‑world settings.

By Jani Nyk\"anen, Otto Tabell, Santtu Tikka, Juha Karvanen
More like this →
arXiv AI
Aug 26

Partial Identification under Causal Orders by Linear Programming

The paper presents a method for partially identifying counterfactual queries without requiring a fully specified causal graph. By exploiting the topological ordering implied by the query itself, the authors transform the identification problem into a linear programming task, enabling bounds on arbitrary counterfactual and nested counterfactual queries. They demonstrate the tightness of these bounds and illustrate the approach on several case studies, showing its usefulness even when causal knowledge is incomplete.

By Eric Rossetto, Alessandro Antonucci
More like this →