Probabilistic Neural Programs

Kenton Murray, Jayant Krishnamurthy

Abstract: We present probabilistic neural programs, a framework for program induction that permits flexible specification of both a computational model and inference algorithm while simultaneously enabling the use of deep neural networks. Probabilistic neural programs combine a computation graph for specifying a neural network with an operator for weighted nondeterministic choice. Thus, a program describes both a collection of decisions as well as the neural network architecture used to make each one. We evaluate our approach on a challenging diagram question answering task where probabilistic neural programs correctly execute nearly twice as many programs as a baseline model.

TL;DR: We present probabilistic neural programs, a framework for program induction using neural nets and probabilistic programming.

Conflicts: nd.ed, allenai.org, cmu.edu