An Introduction to Deep Learning for Vision

Rob Fergus
New York University
Canadian Institute for Advanced Research

Deep Learning is an exciting area of machine learning, which has produced a number of breakthrough results recently in a number of domains such as vision, speech and NLP. This talk will give a high level overview of these methods as applied to object recognition problems. The talk will start by motivating the need to learn features, rather than hand-craft them. It will then introduce several basic architectures, explaining how they learn features, and showing how they can be "stacked" into hierarchies that can extract multiple layers of representation. Throughout, links will be drawn between these methods and existing approaches to recognition, particularly those involving hierarchical representations. The final part of the lecture will examine the current performances obtained by feature learning approaches on a range of standard vision benchmarks, highlighting their strengths and weaknesses.

Presentation (PDF File)

Back to Graduate Summer School: Computer Vision