Dr. Ang Li’s Research Group Has Two Papers Accepted at Top AI Conferences: IJCAI 2026 and AISTATS 2026

Dr. Ang Li

Dr. Ang Li’s research group lands two papers at top AI venues

Department of Computer Science

Dr. Ang Li’s research group in the Department of Computer Science has recently had two papers accepted at leading international conferences in Artificial Intelligence: the International Joint Conference on Artificial Intelligence (IJCAI 2026) and the International Conference on Artificial Intelligence and Statistics (AISTATS 2026).

IJCAI 2026

Identification of Probabilities of Causation: From Recursive to Closed-Form Bounds

This paper advances the theoretical foundations of counterfactual reasoning by establishing a comprehensive framework for probabilities of causation in discrete settings. The authors define four representative probabilities of causation, derive their corresponding bounds, and prove that all other discrete probabilities of causation can be expressed in terms of these four canonical forms. The work represents a significant milestone in the theory of counterfactual analysis and causal inference. By providing a unified mathematical foundation for probabilities of causation, it supports more solid personalized decision-making across domains such as healthcare, education, and public policy.

AISTATS 2026

Epsilon-Identifiability of Causal Quantities

This paper introduces the concept of epsilon-identifiability, extending the classical notion of identifiability in Structural Causal Models developed by Judea Pearl. Because many causal and counterfactual quantities are not fully identifiable from available data, the authors propose a more practical framework in which a causal quantity is considered identifiable within a small error tolerance, denoted by Epsilon. Under this new definition, the paper establishes a broad class of epsilon-identifiability results for causal effects and counterfactual quantities. The work opens a new research direction in causal inference and Structural Causal Models by providing decision-makers with more opportunities to apply causal reasoning when exact identification is impossible but high-precision approximation is achievable.

Learn more about Dr. Li’s research →