Search papers, labs, and topics across Lattice.
CausalSteward (CAST) is a novel human-in-the-loop framework designed to enhance causal discovery from high-dimensional data by employing a multi-agent collaborative system that utilizes a divide-and-conquer strategy. The framework integrates vast amounts of prior knowledge with data-driven methods, leveraging tools like retrieval augmented generation and conditional independence tests to iteratively analyze clusters of variables. Key findings reveal that CAST not only improves the accuracy of causal models but also highlights the critical role of human interaction in refining causal reasoning processes.
A multi-agent framework that combines prior knowledge with data-driven analysis can significantly enhance causal discovery in complex, high-dimensional datasets.
Learning causal models from high-dimensional data is a significant challenge, particularly in real-world settings where violations of core assumptions lead to causal identifiability issues. Although massive amounts of prior knowledge are available, and contain valuable causal information, effectively integrating this knowledge into the causal discovery process remains an open problem. We introduce CausalSTeward (CAST), a novel human-in-the-loop framework for interactively assembling large causal models. CausalSteward is a multi-agent collaborative system that tackles high-dimensional causality through a divide-and-conquer approach where large clusters of variables are iteratively partitioned and then separately analyzed. Our framework fuses prior knowledge with a data-driven approach by using tailored tools such as retrieval augmented generation and conditional independence tests. Finally, we use this work to examine the capabilities and limitations of causal reasoning in multi-agent frameworks, and how the human-in-the-loop can contribute to accurate and trustworthy results.
