Coinductive Trees for Exact Inference of Probabilistic Programs
We propose a coinductive variant of Knuth-Yao trees (variously, Discrete Distribution Generating trees) as an intermediate representation supporting exact inference for probabilistic programs which may contain loops with probabilistic termination conditions. We provide a prototype implementation of a probabilistic programming language, Zar, written in Haskell, exemplifying both Knuth-Yao trees as well as exact inference on the trees.
Tue 21 Jan Times are displayed in time zone: Saskatchewan, Central America change
15:35 - 17:45
|Coinductive Trees for Exact Inference of Probabilistic Programs|
|Name generation and Higher-order Probabilistic Programming (Or is new=rnd?)|
|Density Functions of Statistical Probabilistic Programs|
|Probabilistic Programming around Gaussian Processes|