Workshop @ UAI 2026
August 21st, 2026, Amsterdam, NL
Modern machine learning systems increasingly operate in interactive environments where decisions influence future data. While recent advances in deep learning have driven remarkable progress across domains such as natural language processing, computer vision, and decision-making, a central challenge remains: achieving reliable generalization beyond the i.i.d. setting, especially under distribution shifts and in rapidly changing environments. Addressing this challenge requires moving beyond purely statistical associations toward understanding the underlying causal structure of the world.
Causal modeling provides a principled framework for representing mechanisms, reasoning about interventions, and evaluating counterfactual outcomes. By combining data with structural assumptions about the environment, causal inference enables reasoning about causal effects. This perspective is especially important in decision-making frameworks such as reinforcement learning (RL), offline policy evaluation, planning under uncertainty, sequential decision-making, and multi-armed bandits (MABs). These settings require not only accurate prediction, but also an understanding of how actions shape future outcomes. Incorporating causal knowledge into decision-making systems can improve predictions about the consequences of actions across varying conditions, enable more accurate diagnosis for better policy selection, and support more efficient exploration.
Despite this natural complementarity, causal inference and decision-making research have largely developed in parallel, with limited interaction. Yet both ultimately address different aspects of the same core problem: understanding and exploiting factual and counterfactual relationships in complex environments.
The UAI community has played a central role in advancing both causal inference and decision-theoretic reasoning. This workshop aims to bridge these two pillars by providing a focused venue for exploring how causal reasoning can enhance decision-making algorithms, particularly in reinforcement learning and related paradigms. The various sub-communities within the causal inference and decision making (CIDM) spectrum regularly publish at leading machine learning venues such as ICML, NeurIPS, and UAI. This year’s workshop seeks to bring together researchers across these areas, as well as those beyond the immediate scope of CIDM. A diverse lineup of distinguished invited speakers, together with an open discussion panel, will foster the exchange of ideas and help disseminate recent advances not only within the CIDM community but also to the broader UAI audience.
Keynote Speakers
This workshop builds on a growing interest in integrating causality with decision-making and reinforcement learning. Related workshops include:
Causal Learning for Decision Making, ICLR 2020.
Causality: Learning, Inference, and Decision-Making, UAI 2017.
Workshops on causal representation learning and causal RL at major AI conferences.
Since the last dedicated workshop on causal learning for decision making, the area has undergone theoretical and empirical growth. Advances in offline reinforcement learning, causal representation learning, and robustness under distribution shift have brought causal reasoning to the forefront of modern decision-making research. The field has reached a stage where unifying currently fragmented lines of research and establishing clearer theoretical foundations is both timely and necessary. UAI, with its longstanding strengths in causality and decision theory, provides a natural venue to articulate shared principles, clarify open challenges, and guide the next directions of research at the intersection of causality and sequential decision-making.
Venue