This short course will come in two parts. In the first part, we will present a general model of decision forests and discuss how it can be used for a large variety of supervised and unsupervised tasks in machine learning and computer vision. Numerous toy examples will help explain and demonstrate how small variants of the basic forest model correspond to powerful algorithms for efficient classification, regression, density estimation, manifold learning and semi-supervised learning. We show how these techniques can be applied to real-world applications including human tracking in Microsoft Kinect. The second part will then explain how random field models can be built upon per-pixel decision forests to provide spatial priors, leading to the decision tree field and regression tree field models.
The tutorial has been recorded and the videos are available at TechTalks.TV.
The tutorial will take place on the 2nd December 2013 between 9am and 1pm in room 202. The schedule below is preliminary and there may be some last minute changes before the tutorial takes place.
|9:00-10:30||Randomized Decision Forests and their Applications in Computer Vision (Jamie), PPT|
|10:30-10:50||Entropy estimation and streaming data (Sebastian), PPT|
|10:50-11:30||Decision Jungles (Jamie), second half of above PPT|
|11:30-13:00||Decision and Regression Tree Fields (Sebastian), PPT, and some material from PDF|
In addition to the material linked above in the schedule, take a look at the book page.