The intersection of artificial intelligence, machine learning and statistics. He is particularly interested in hierarchical, probabilistic graphical models and approximate inference and learning techniques including Markov Chain Monte Carlo and variational Bayesian methods. He is also interested in non-likelihood-based estimation for intractable probabilistic models including Markov Random Fields and Restricted Boltzmann Machines. Professor Marlin works on a broad range of applications for these modeling and learning techniques including collaborative filtering and recommender systems, ranking, unsupervised structure discovery and feature induction, object recognition and image labeling, and medical informatics.
