We consider the problem of dynamically apportioning resources among a set of options in a worst-case on-line framework. The model we study can be interpreted as a broad, abstract extension of the well-studied on-line prediction model to a general decision-theoretic setting. We show that the multiplicative weight-update rule of Littlestone and Warmuth 10 can be adapted to this mode yielding bounds that are slightly weaker in some cases, but applicable to a considerably more general class of learning problems. We show how the resulting learning algorithm can be applied to a variety of problems, including gambling, multiple-outcome prediction, repeated games and prediction of points in ℝ n . We also show how the weight-update rule can be used to derive a new boosting algorithm which does not require prior knowledge about the performance of the weak learning algorithm.
Description
A desicion-theoretic generalization of on-line learning and an application to boosting | SpringerLink
%0 Book Section
%1 Freund1995adaboost
%A Freund, Yoav
%A Schapire, Robert E.
%B Computational Learning Theory: Second European Conference, EuroCOLT '95 Barcelona, Spain, March 13--15, 1995 Proceedings
%C Berlin, Heidelberg
%D 1995
%I Springer Berlin Heidelberg
%K lursurvey
%P 23--37
%R 10.1007/3-540-59119-2_166
%T A desicion-theoretic generalization of on-line learning and an application to boosting
%U https://doi.org/10.1007/3-540-59119-2_166
%X We consider the problem of dynamically apportioning resources among a set of options in a worst-case on-line framework. The model we study can be interpreted as a broad, abstract extension of the well-studied on-line prediction model to a general decision-theoretic setting. We show that the multiplicative weight-update rule of Littlestone and Warmuth 10 can be adapted to this mode yielding bounds that are slightly weaker in some cases, but applicable to a considerably more general class of learning problems. We show how the resulting learning algorithm can be applied to a variety of problems, including gambling, multiple-outcome prediction, repeated games and prediction of points in ℝ n . We also show how the weight-update rule can be used to derive a new boosting algorithm which does not require prior knowledge about the performance of the weak learning algorithm.
%@ 978-3-540-49195-8
@inbook{Freund1995adaboost,
abstract = {We consider the problem of dynamically apportioning resources among a set of options in a worst-case on-line framework. The model we study can be interpreted as a broad, abstract extension of the well-studied on-line prediction model to a general decision-theoretic setting. We show that the multiplicative weight-update rule of Littlestone and Warmuth [10] can be adapted to this mode yielding bounds that are slightly weaker in some cases, but applicable to a considerably more general class of learning problems. We show how the resulting learning algorithm can be applied to a variety of problems, including gambling, multiple-outcome prediction, repeated games and prediction of points in ℝ n . We also show how the weight-update rule can be used to derive a new boosting algorithm which does not require prior knowledge about the performance of the weak learning algorithm.},
added-at = {2017-12-30T18:34:14.000+0100},
address = {Berlin, Heidelberg},
author = {Freund, Yoav and Schapire, Robert E.},
biburl = {https://www.bibsonomy.org/bibtex/239dc487a98c90caa6e46ebcabe4f1732/lautenschlager},
booktitle = {Computational Learning Theory: Second European Conference, EuroCOLT '95 Barcelona, Spain, March 13--15, 1995 Proceedings},
description = {A desicion-theoretic generalization of on-line learning and an application to boosting | SpringerLink},
doi = {10.1007/3-540-59119-2_166},
interhash = {980905f284b39dedbb5ff6299276f076},
intrahash = {39dc487a98c90caa6e46ebcabe4f1732},
isbn = {978-3-540-49195-8},
keywords = {lursurvey},
pages = {23--37},
publisher = {Springer Berlin Heidelberg},
timestamp = {2018-01-19T12:16:18.000+0100},
title = {A desicion-theoretic generalization of on-line learning and an application to boosting},
url = {https://doi.org/10.1007/3-540-59119-2_166},
year = 1995
}