My work has focused on probabilistic modeling of high-dimensional data, and the application of this technique to various domains. Specifically, I approach data, such as real images and feasible robot trajectories, as samples drawn from complex distributions. I investigate the representation, learning, and sampling of the underlying distributions.