Fast Exact Bayesian Inference for High-Dimensional Models

Abstract

In this text, we present the principles that allow the tractable implementation of exact inference processes concerning a group of widespread classes of Bayesian generative models, which have until recently been deemed as intractable whenever formulated using high-dimensional joint distributions. We will demonstrate the usefulness of such a principled approach with an example of real-time OpenCL implementation using GPUs of a full-fledged, computer vision-based model to estimate gaze direction in human-robot interaction (HRI).