We introduce NN-GRISE, a neural net based algorithm to learn a graphical model given i.i.d samples from its joint probability distribution. This algorithm minimizes an interaction screening objective function to learn neural network representations of conditionals of a graphical model. The NN-GRISE algorithm is a good alternative to traditional graphical model learning techniques when the true graphical model has higher-order interactions with a high degree of symmetry. In these cases, NN-GRISE is able to find the correct parsimonious representation for the conditionals without being fed any prior information about the model. We also show how this method can be used to learn the structure of a graphical model.
