TY - GEN
T1 - Ensemble methods for structured prediction
AU - Cortes, Corinna
AU - Kuznetsov, Vitaly
AU - Mohri, Mehryar
N1 - Funding Information:
Financial support was provided by The Petroleum Research Fund (administered by the American Chemical Society), the Geological Society of America, White Mountain Research Station, and the Wilbur Sherman Fellowship Fund (UCLA). We thank the following for help and advice at various stages of this project: Bruce Bilodeau, Peter Bird, Dave Bloom, Tim Boyle, William Cavazza, Kevin Corbett, Greg Davis, Andrew Diamond, Deborah Diamond, Rita Diamond, Tom Drake, Brad Hacker, Bill Heins, Dave Kemp, Kathie Marsaglia, Clem Nelson, Gerhard Oertel, Everett Olson, Tony Orme, Thor Riksheim, Floyd Sabins, Jack Stewart, and An Yin. We also thank Peter Bird, Everett Olson, Tony Orme, and An Yin for reviewing earlier versions of this manuscript. We especially thank Marty Grove for help with Ar-Ar analysis and preparation of Figure 4, and An Yin and Gary Ernst for reviewing the final manuscript. Bruce, Chris, Mike Sweet, Stubby, the ladies at Gramma Read’s store, and the populace of Silver Peak, in general, provided invaluable hospitality and companionship during our visits there.
Publisher Copyright:
Copyright © (2014) by the International Machine Learning Society (IMLS) All rights reserved.
PY - 2014
Y1 - 2014
N2 - We present a series of learning algorithms and theoretical guarantees for designing accurate en-sembles of structured prediction tasks. This includes several randomized and deterministic algorithms devised by converting on-line learning algorithms to batch ones, and a boosting-style algorithm applicable in the context of structured prediction with a large number of labels. We give a detailed study of all these algorithms, including the description of new on-line-to-batch conversions and learning guarantees. We also report the results of extensive experiments with these algorithms in several structured prediction tasks.
AB - We present a series of learning algorithms and theoretical guarantees for designing accurate en-sembles of structured prediction tasks. This includes several randomized and deterministic algorithms devised by converting on-line learning algorithms to batch ones, and a boosting-style algorithm applicable in the context of structured prediction with a large number of labels. We give a detailed study of all these algorithms, including the description of new on-line-to-batch conversions and learning guarantees. We also report the results of extensive experiments with these algorithms in several structured prediction tasks.
UR - http://www.scopus.com/inward/record.url?scp=84919808185&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84919808185&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:84919808185
T3 - 31st International Conference on Machine Learning, ICML 2014
SP - 2856
EP - 2872
BT - 31st International Conference on Machine Learning, ICML 2014
PB - International Machine Learning Society (IMLS)
T2 - 31st International Conference on Machine Learning, ICML 2014
Y2 - 21 June 2014 through 26 June 2014
ER -