causal discovery

Joint Causal Inference from Multiple Contexts

We show that one can reduce a causal discovery problem on several different observational and experimental settings to a problem on a single distribution with the addition of several context variables. We call this approach Joint Causal Inference and show it can be combined with many algorithms, e.g. also algorithms that can deal with cycles, latent confounders and selection bias.

Joris M. Mooij, Sara Magliacane, Tom Claassen

Active Structure Learning of Causal DAGs via Directed Clique Trees

A growing body of work has begun to study intervention design for efficient structure learning of causal directed acyclic graphs …

Chandler Squires, Sara Magliacane, Kristjan Greenewald, Dmitriy Katz, Murat Kocaoglu, Karthikeyan Shanmugam

Sample Efficient Active Learning of Causal Trees

We consider the problem of experimental design for learning causal graphs that have a tree structure. We propose an adaptive framework …

Kristjan Greenewald, Dmitriy Katz, Karthikeyan Shanmugam, Sara Magliacane, Murat Kocaoglu, Enric Boix Adsera, Guy Bresler