Static Posterior Inference of Bayesian Probabilistic Programming via Polynomial Solving

Explore a groundbreaking automated approach for deriving guaranteed bounds on normalized posterior distributions in probabilistic programming through polynomial solving. Delve into the innovative method that handles programs with unbounded while loops and continuous distributions with infinite supports. Learn about the classification of programs into 'score-at-end' and 'score-recursive' categories, and understand how a fixed-point theorem and a multiplicative variant of the Optional Stopping Theorem are applied to infer bounds on normalized posterior distributions. Gain insights from the research presented by Peixin Wang, Hongfei Fu, Tengshun Yang, Guanyan Li, and C.-H. Luke Ong in this 11-minute conference talk from ACM SIGPLAN's LAFI'24 event.