![[2602.14503] Bounding Probabilities of Causation with Partial Causal Diagrams](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.14503] Bounding Probabilities of Causation with Partial Causal Diagrams
Summary
This paper presents a framework for bounding probabilities of causation using partial causal diagrams, addressing limitations of existing methods that require complete causal graphs or binary settings.
Why It Matters
Understanding probabilities of causation is crucial for decision-making and explanations in AI. This research expands the applicability of causal inference methods to real-world scenarios where causal knowledge is often incomplete, enhancing the robustness of AI systems.
Key Takeaways
- Proposes a new framework for bounding probabilities of causation.
- Incorporates partial causal information into optimization formulations.
- Addresses limitations of existing methods that require complete causal graphs.
- Enhances decision-making capabilities in AI applications.
- Validates tighter bounds without needing full identifiability.
Computer Science > Artificial Intelligence arXiv:2602.14503 (cs) [Submitted on 16 Feb 2026] Title:Bounding Probabilities of Causation with Partial Causal Diagrams Authors:Yuxuan Xie, Ang Li View a PDF of the paper titled Bounding Probabilities of Causation with Partial Causal Diagrams, by Yuxuan Xie and Ang Li View PDF HTML (experimental) Abstract:Probabilities of causation are fundamental to individual-level explanation and decision making, yet they are inherently counterfactual and not point-identifiable from data in general. Existing bounds either disregard available covariates, require complete causal graphs, or rely on restrictive binary settings, limiting their practical use. In real-world applications, causal information is often partial but nontrivial. This paper proposes a general framework for bounding probabilities of causation using partial causal information. We show how the available structural or statistical information can be systematically incorporated as constraints in a optimization programming formulation, yielding tighter and formally valid bounds without full identifiability. This approach extends the applicability of probabilities of causation to realistic settings where causal knowledge is incomplete but informative. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2602.14503 [cs.AI] (or arXiv:2602.14503v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2602.14503 Focus to learn more arXiv-issued DOI via DataCite (pending regist...