About cookies on this site Our websites require some cookies to function properly (required). In addition, other cookies may be used with your consent to analyze site usage, improve the user experience and for advertising. For more information, please review your options. By visiting our website, you agree to our processing of information as described in IBM’sprivacy statement. To provide a smooth navigation, your cookie preferences will be shared across the IBM web domains listed here.
Publication
ICML 2022
Poster
Entropic Causal Inference: Graph Identifiabilty
Abstract
Entropic causal inference is a recent framework for learningthe causal graph between two variables from observationaldata for structural causal models with small entropy. In thispaper, we first extend the causal graph identifiability resultin the two-variable setting under relaxed assumptions. Next,we show the first identifiability result using the entropic ap-proach for learning causal graphs with more than two nodes.We provide a sequential peeling algorithm that is provably cor-rect for general graphs. We also propose a heuristic algorithmfor small graphs that shows strong empirical performance.We conduct rigorous experiments that demonstrate the perfor-mance of our algorithms compared to the existing work usingsynthetic data with different generative models. Finally wetest our algorithms on real-world datasets.