Recent Talks

• Boosting and Predictive modeling (PowerPoint presentation, 4.7MB)
  
  
• A black-box approach to machine learning (PowerPoint presentation, 1MB)
  
  
• How to be a Bayesian without believing  (PowerPoint presentation, 908KB)
  A talk about the paper  Generalization bounds for averaged classifiers
  Presented in Eurandom, Eindhoven, the Netherlands, January 2003.
  Slide no. 33 in this presentation is a link to the following mpeg movie (13.2MB)
  created by Paul Viola and Mike Jones which demonstrates their amazing face detection software.
  
  
• An Introduction to boosting  (PowerPoint presentation, 688KB)
  An introductory talk about boosting, margins, and alternating trees.
  
  
• Games of Prediction  (PowerPoint presentation, 560KB)
  A talk which shows how algorithms for boosting, online learning, and Ulam's game with lies 
  can all be seen as special cases of interactive games Schapire named "drifting games".
  
  
• Why does averaging work?  (PowerPoint presentation, 473KB)
  A talk given in the conference Interface 2001. 
  Introduces some of the new ideas which machine learning brings to statistical inference, especially regarding
  classification methods that use averaging.
  
  
• Computational and statistical issues in data-mining  (PowerPoint presentation, 1175KB)
  A talk to be presented  in SAS M2002 conference on data-mining.