A hybrid approach for large scale causality discovery

Abstract

Causality discovery is one of the basic problems in the scientific field. Though many researchers are committed to find the causal relation from observational data, there are still no effective methods for the high dimensional data. In this work, we propose a hybrid approach by taking the advantage of two state of the art causal discovery methods. In the proposed method, the structure learning based methods are explored to discover the causal skeleton, and then the additive noise models are conducted to distinguish the direction of causalities. The experimental results show that the proposed approach is effective and scalable for the large scale causality discovery problems. © Springer-Verlag Berlin Heidelberg 2013.