Topics in Systems: Learning by Boosting and Adaptation: AdaBoost presented by P.R. Kumar

Topics in Systems: Learning by Boosting and A…

Topics in Systems: Learning by Boosting and A…
Viacheslav Fedorov presents Area Efficient TCAM design

Viacheslav Fedorov presents Area Efficient TC…

Viacheslav Fedorov presents Area Efficient TC…
Andrew Targhetta Defends Thesis on Ultra-low Energy Asymmetric Cryptography

Andrew Targhetta Defends Thesis on Ultra-low …

Andrew Targhetta Defends Thesis on Ultra-low …
B-Fetch Branch Prediction Paper nominated for best paper at MICRO conference

B-Fetch Branch Prediction Paper nominated for…

B-Fetch Branch Prediction Paper nominated for…
Zhang Receives Best Paper Award at IEEE GLOBECOM 2014

Zhang Receives Best Paper Award at IEEE GLOBE…

Zhang Receives Best Paper Award at IEEE GLOBE…
PreviousNext

Topics in Systems: Learning by Boosting and Adaptation: AdaBoost presented by P.R. Kumar

Last week, Dr. P.R. Kumar provided an account of the key papers by Yoav Freund and Robert E. Schapire that developed AdaBoost, a major milestone in machine learning. This seminar attracted a large audience, comprised of both students and faculty. Adaptive Boosting, or “AdaBoost” for short, is a machine learning meta-algorithm. This particular algorithm is used in conjunction with others to improve their performance. Freund and Schapire won the Gödel Prize in 2003 for their work.

 

The paper describes a primary algorithm interpreted as an extension of the on-line prediction model to a general decision-theoretic setting. This algorithm is shown to be applicable to a wider, more general class of machine-learning problems. These applications include problems like gambling, repeated games, and multiple-outcome prediction. Then a boosting algorithm, derived from the multiplicative weight-update technique, is presented. Finally, the boosting algorithm is studied to show it does not need prior knowledge of other, weak learning algorithms.

 

Dr. P.R. Kumar obtained his Bachelor of Technology degree in Electrical Engineering from I.I.T Madras in 1973, and his M.S. and D.Sc. degrees in Systems Science and Mathematics from Washington University, St. Louis, in 1975 and 1977. Currently he holds the College of Engineering Chair in Computer Engineering at Texas A&M. Kumar has worked on problems in wide range, including game theory and machine-learning. He is a member of the National Academy of Engineering of the USA, and a Fellow of the Academy of Sciences of the Developing World.