Monte Carlo Semantic Differencing of Probabilistic Programs
This talk introduces a Monte Carlo dynamic analysis of two probabilistic programs. The analysis estimates an upper bound on the difference in their output distributions for ﬁxed inputs, as measured by Kullback-Leibler (KL) divergence. The analysis, Bridged Auxiliary Inference Divergence Estimator (BRAIDE), is a generalization of the Auxiliary Inference Divergence Estimator (AIDE). Unlike AIDE, BRAIDE analysis can be made more precise and efﬁcient using knowledge of how the traces of the two programs relate, resulting in tighter bounds with less computation. We give an example of BRAIDE applied to two Gen probabilistic programs.
Tue 21 JanDisplayed time zone: Saskatchewan, Central America change
14:00 - 15:05
|Divide, Conquer, and Combine: a New Inference Strategy for Probabilistic Programs with Stochastic Support|
|MetaPPL: Inference Algorithms as First-Class Generative Models|
Alexander K. Lew Massachusetts Institute of Technology, USA, Benjamin Sherman Massachusetts Institute of Technology, USA, Marco Cusumano-Towner MIT-CSAIL, Austin Garrett MIT, Ben Zinberg MIT, Vikash K. Mansinghka MIT, Michael Carbin Massachusetts Institute of TechnologyFile Attached
|Monte Carlo Semantic Differencing of Probabilistic Programs|