I develop novel algorithms and models for artificial intelligence and machine
My current research interest is in the following areas:
1. probabilistic deep learning methods for unsupervised learning and
semi-supervised structured prediction;
2. the consequence of model misspecification on estimation and on decision
3. understanding agent complexity in order to improve learning efficiency;
4. designing models for reasoning and planning.
In terms of methodology I am interested in structured prediction, deep
learning, Monte Carlo methods, and decision theory.
My work has found industrial applications in computer vision, cloud-based
machine learning, gesture recognition, statistical image processing,
numerical optimization, and in time-of-flight imaging.