Program Synthesis: A Hierarchical Nonparametric Bayesian Approach
Speaker: Percy Liang , UC Berkeley
Date: March 31 2010
Time: 4:00PM to 5:00AM
Location: 32-G449
Host: Josh Tenenbaum, BCS, MIT
Contact: Ruslan Salakhutdinov, rsalakhu@mit.edu
Relevant URL: In this talk, we focus on the problem of inferring computer programs from their behavior on one or two training examples. A single program is likely to be undetermined by this data. Therefore, we introduce a nonparametric hierarchical Bayesian prior over programs which allows the sharing of statistical strength across multiple programs. The key challenge we address is how to parametrize this multi-task sharing. For this, we introduce a new representation of programs based on combinatory logic and provide an MCMC algorithm that can perform safe program transformations on this representation in order to reveal shared inter-program substructures. We show the benefits of our multi-task approach on learning macros in a text-editing domain.
Bio
Percy Liang is currently a Ph.D. student at UC Berkeley with Michael Jordan and Dan Klein, having completed his MEng at MIT with Michael Collins in 2005. His current research focuses on grounded natural language semantics, in particular, using programs and probabilistic modeling to represent and learn associations between linguistic units and the non-linguistic world. His other work includes developing efficient algorithms for learning latent-variable models and theoretical analyses of generative/discriminative learning and regularization.
See other events that are part of
See other events happening in March 2010
