An Approach for Inferring Causal Directions from Multi-Dimensional Networks

Inferring causal directions from observed variables is one of the fundamental problems in many scientific fields. In this paper, a new approach for causal-direction inference from multi-dimensional networks is proposed based on a split-and-merge strategy. The method first decomposes an n-dimensional network into induced subnetworks, each of which corresponds to a node in the network. It shows that each induced subnetwork can be subsumed to one of the three substructures: one-degree, non-triangle and triangle-existence substructures. Three effective algorithms are developed to infer causalities from the three substructures. The whole causal structure of the multi-dimensional network is obtained by learning these induced subnetworks separately. Experimental results demonstrate that our method is more general and effective than the state-of-the-art methods.