General info: Machine Learning for Computer Graphics
Time: MW 3-4:30
Location: SS 1080
Instructor: Aaron Hertzmann (hertzman@dgp.toronto.edu)
Office: BA 5268
Office hours: drop in or by appointment
Web page: www.cs.toronto.edu/~csc2521h
Newsgroup: ut.cdf.csc2521h
Textbook: Information Theory, Inference, and Learning Algorithms, by David MacKay. This book is in the bookstore, and is also available for onscreen viewing online.
Sophisticated computer graphics applications require complex models of appearance, human motion, natural phenomena, and even artistic style. Such models are often difficult or impossible to design by hand. Recent research demonstrates that, instead, we can "learn" a dynamical and/or appearance model from captured data, and then use the model to synthesize plausible new data. For example, we can capture the motions of a human actor, and then generate new motions as they might be performed by that actor.
In this course, we will survey basic principles of machine learning, and how they can be applied to real problems in computer graphics and animation. Bayesian methods will be emphasized; most topics will also relate to problems in computer vision. The format will be a mix of lectures, student paper presentations, and discussion. The final projects will be research-oriented, intended to explore new areas of this emerging field, and ultimately lead to projects and publications in leading graphics, vision, and learning conferences.
CS grads or instructor permission; familiarity with probability, statistics, and linear algebra required (although you do not need to be an expert). Experience with computer graphics, vision, or learning useful but not required.