Bayesian Nets from the ground up

Abstract

In this tutorial Aish will take us through Bayesian Networks (i.e. Directed Graphical Models) from the ground up. Many real-world problems in machine learning benefit from building custom models and explicitly stating your distributional assumptions. Graphical models provide a general methodology for doing this. They’ve found success in such diverse settings as bioinformatics, speech processing, and driving parts of Netflix’s recommendation engine.

Aish will start from the basics and build up to more advanced concepts, such as bayesian nonparametric extensions. By the end of the tutorial you should have a gasp on the theory underpinning Bayes nets, how to build your own models, and how to infer them.

Bio

Aish manages machine learning research at Netflix and leads an applied research team working on the core recommender system. His work combines cutting edge machine learning with large scale software engineering. Prior to Netflix, Aish was head of data science at Opentable, and also founded his own startup, backed by Sam Morgan, for solving large-scale operations research problems.

Resources