A probabilistic autoencoder for causal discovery

Matthias Feiler

The paper addresses the problem of finding the causal direction between two associated variables. The proposed solution is to build an autoencoder of their joint distribution and t…