Beyond Bayesian Model Averaging over Paths in Probabilistic Programs with Stochastic Support

arXiv:2310.14888v2 Announce Type: replace
Abstract: The posterior in probabilistic programs with stochastic support decomposes as a weighted sum of the local posterior distributions associated with each possible program path. We show that making predictions with this full posterior implicitly performs a Bayesian model averaging (BMA) over paths. This is potentially problematic, as BMA weights can be unstable due to model misspecification or inference approximations, leading to sub-optimal predictions in turn. To remedy this issue, we propose alternative mechanisms for path …

abstract arxiv bayesian beyond cs.lg cs.pl making path posterior predictions show stochastic sum support type