Boston Lisp April 27, 2009

Noah Goodman on Lambda the Ultimate Gamble

Where: MIT 34-401B

MIT map: http://whereis.mit.edu/bin/map?selection=34
Google map: http://maps.google.com/maps?q=50+Vassar+St,+Cambridge,+MA+02139,+USA

Noah Goodman will talk about Church: a language for probabilistic modeling, or, Lambda, the Ultimate Gamble. He will describe the probabilistic programming language Church. Probabilistic generative models have exploded in recent years, becoming central to machine learning and AI. These models are usually described with a mixture of informal english, math, and box-and-arrow diagrams. Such descriptions can be error prone and are difficult to scale in model complexity. Church is a formal language for probabilistic generative models, derived from the pure subset of Scheme and extended with probabilistic constructs. As a description language Church is a convenient and powerful way to construct models; in this talk I will show several examples drawn from recent machine learning research. Beyond mere description, Church makes it possible to automate the process of inference in probabilistic models. The MIT-Church implementation of Church is a universal inference engine based on Markov chain monte carlo. I will indicate the design of this implementation and highlight some of the unique challenges of probabilistic programming languages relative to standard languages. I will close with some examples of MIT-Church in action.

Noah D. Goodman is a research scientist in the Department of Brain and Cognitive Sciences at MIT, and a member of the Computer Science and Artificial Intelligence Laboratory. He studies the computational basis of human thought, merging behavioral experiments with formal methods from statistics and logic. He received
his Ph.D. in mathematics from the University of Texas at Austin, several years later he joined the Computational Cognitive Science group at MIT, working with professor Joshua Tenenbaum. Goodman has published more than twenty-five publications in psychology, cognitive science, and artificial intelligence. Goodman is project leader of the MIT-Church probabilistic programming project.

His website is at http://www.mit.edu/~ndg/

Having observed the success of the formula at ILC'2009, we have instituted Lightning Talks at the Boston Lisp Meeting. At every meeting, before the main talk, there are two slots for strictly timed 5-minute talks followed by 2-minute for questions and answers.

The slots for next Monday are still open. Step up and come talk about your pet project!