Search |  Contact |  SRI Home Do not follow this link, or your host will be blocked from this site. This is a spider trap. Do not follow this link, or your host will be blocked from this site. This is a spider trap. Do not follow this link, or your host will be blocked from this site. This is a spider trap.A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A ASRI International.  333 Ravenswood Avenue.  Menlo Park, CA 94025-3493. SRI International is a nonprofit corporation.

AIC Seminar Series

Functional Symmetries in Probabilistic Programs

Lingfeng YangStanford University[Home Page]

Notice:  Hosted by Rodrigo Braz

Date:  2013-12-12 at 16:00

Location:  EJ228 (SRI E building)  (Directions)

   Abstract

Probabilistic programs can compactly express large distributions through functional abstraction and iteration/recursion. This compactness of description has a further consequence: the random choices generated by running the program contain a great deal of symmetry. This has consequences for the design of inference algorithms. I will focus on MCMC in particular. First, I present the concept of functional symmetry in probabilistic programs, where I build upon the view (of Kiselyov [1]) of user-defined functions as Herbrand universe generators of first-order logic predicates. Two subsets of random choices in the program are functionally symmetric if there is a common function call that generates both. This leads to an MCMC inference algorithm, flow-automorphic structure tempering (FAUST). It operates on the structure of the program traces, swapping these functionally symmetric subsets of random choices. For many problems, the result is much improved mixing. Finally, I conclude with descriptions of further work that this perspective suggests. [1] http://okmij.org/ftp/kakuritu/logic-programming.html#logic-var

   Bio for Lingfeng Yang

Lingfeng Yang is a 6th year Ph. D candidate working at Stanford University with Pat Hanrahan and Noah Goodman. He enjoys investigating symbolic processes, watching stochastic systems work, and driving.

   Note for Visitors to SRI

Please arrive at least 10 minutes early as you will need to sign in by following instructions by the lobby phone at Building E. SRI is located at 333 Ravenswood Avenue in Menlo Park. Visitors may park in the parking lots off Fourth Street. Detailed directions to SRI, as well as maps, are available from the Visiting AIC web page. There are two entrances to SRI International located on Ravenswood Ave. Please check the Builing E entrance signage.

SRI International
©2014 SRI International 333 Ravenswood Avenue, Menlo Park, CA 94025-3493
SRI International is an independent, nonprofit corporation. Privacy policy