Conditional independence testing, two sample comparison and density estimation using neural networks

Abstract

Given the vast amount of data available nowadays and the rapid increase of computational processing power, the field of machine learning and the so called algorithmic modeling have seen a recent surge in its popularity and applicability. One of the tools which has attracted great popularity is artificial neural networks due, to among other things, their versatility, ability to capture complex relations and computational scalability. In this work, we therefore apply such machine learning tools into three important problems of Statistics: two-sample comparison, conditional independence testing and conditional density estimation.