|Stochastic Systems Group|
A Reinforcement Learning Approach to Variational Inference in Probabilistic Programs
We adopt a dynamical systems perspective on variational inference in deep generative models. This connects variational inference to a temporal credit assignment problem that can be solved using reinforcement learning: policy search methods (such as policy gradients) become a direct search through variational parameters; state-space estimation becomes structured variational inference, and temporal-difference methods suggest novel inference algorithms. I will illustrate the technique on structured models from geological modeling.
Problems with this site should be emailed to firstname.lastname@example.org