Abstract

I consider two problems in machine learning and statistics: the problem ofestimating the joint probability density of a collection of random variables,known as density estimation, and the problem of inferring model parameters whentheir likelihood is intractable, known as likelihood-free inference. Thecontribution of the thesis is a set of new methods for addressing theseproblems that are based on recent advances in neural networks and deeplearning.

 

Quick Read (beta)