Universal Marginaliser for Deep Amortised Inference for Probabilistic Programs
This addresses the problem of efficient and reliable inference in probabilistic programming for researchers and practitioners, though it appears incremental as it builds on existing amortised inference techniques.
The paper tackles the computational cost and lack of theoretical guarantees in inference methods for probabilistic programming languages by introducing the Universal Marginaliser, a novel amortised inference method that trains one neural network to approximate any conditional marginal distribution using samples from the prior. The method is benchmarked on multiple probabilistic programs in Pyro with different model structures.
Probabilistic programming languages (PPLs) are powerful modelling tools which allow to formalise our knowledge about the world and reason about its inherent uncertainty. Inference methods used in PPL can be computationally costly due to significant time burden and/or storage requirements; or they can lack theoretical guarantees of convergence and accuracy when applied to large scale graphical models. To this end, we present the Universal Marginaliser (UM), a novel method for amortised inference, in PPL. We show how combining samples drawn from the original probabilistic program prior with an appropriate augmentation method allows us to train one neural network to approximate any of the corresponding conditional marginal distributions, with any separation into latent and observed variables, and thus amortise the cost of inference. Finally, we benchmark the method on multiple probabilistic programs, in Pyro, with different model structure.