Introduction to Graphical Models

Tom Griffiths
University of California, Berkeley (UC Berkeley)

Defining and working with probabilistic models presents a number of challenges, both conceptual and computational. Recent work in computer science and statistics has provided a variety of tools for addressing these challenges, one of which is the language of graphical models.
Graphical models represent probability distributions using a simple graph-based semantics, making it easy to represent and reason about complex, structured probability distributions. I will outline the basic ideas behind graphical models, illustrating how they can be used to support sophisticated probabilistic models of cognition.

Presentation (PowerPoint File)

Back to Graduate Summer School: Probabilistic Models of Cognition: The Mathematics of Mind