The theoretical machine learning group at Princeton invents fundamental new techniques for machine learning and artificial intelligence. Currently many practical successes of machine learning rely upon algorithms that lack guarantees on running time or quality of the solution produced. We seek to rectify this state of affairs through new models, new algorithms and new modes of analysis.
Our research draws from mathematical optimization (convex and nonconvex), statistics, game theory, natural language processing, reinforcement learning, etc. This group also has strong connections to theoretical computer science, complexity theory, game theory and approximation algorithms.
Princeton has always been a wonderful place for fundamental research, and was home to Turing, von Neumann and others who fundamentally contributed to the study of artificial intelligence.