Ultraconservative Online Algorithms for Multiclass Problems

by: Koby Crammer, Yoram Singer

Vol. 2111 (2001), pp. 99-115.

View FullText article

CiteSeer (at Penn State), CiteSeer (at MIT), CiteSeer (at Zurich), CiteSeer (at Singapore)

Reviews [Write a review of this article]

There are no reviews of this article

Find related articles from these CiteULike users

markusd

Find related articles with these CiteULike tags

machine-learning, mira

摘要

In this paper we study a paradigm to generalize online classification algorithms for binary classification problems to multiclass problems. The particular hypotheses we investigate maintain one prototype vector per class. Given an input instance, a multiclass hypothesis computes a similarityscore between each prototype and the input instance and sets the predicted label to be the index of the prototype achieving the highest similarity. To design and analyze the learning algorithms in this...

@inproceedings{citeulike:3015179, abstract = {In this paper we study a paradigm to generalize online classification algorithms for binary classification problems to multiclass problems. The particular hypotheses we investigate maintain one prototype vector per class. Given an input instance, a multiclass hypothesis computes a similarityscore between each prototype and the input instance and sets the predicted label to be the index of the prototype achieving the highest similarity. To design and analyze the learning algorithms in this...}, author = {Crammer, Koby and Singer, Yoram }, booktitle = {14th Annual Conference on Computational Learning Theory, COLT 2001 and 5th {E}uropean Conference on Computational Learning Theory, {EuroCOLT} 2001, Amsterdam, The Netherlands, July 2001, Proceedings}, citeulike-article-id = {3015179}, keywords = {machine-learning, mira}, pages = {99--115}, posted-at = {2008-07-17 18:35:58}, priority = {2}, publisher = {Springer, Berlin}, title = {Ultraconservative Online Algorithms for Multiclass Problems}, url = {http://citeseer.ist.psu.edu/crammer03ultraconservative.html}, volume = {2111}, year = {2001} }

RIS record

TY - CONF ID - citeulike:3015179 L3 - citeulike-article-id:3015179 TI - Ultraconservative Online Algorithms for Multiclass Problems T2 - 14th Annual Conference on Computational Learning Theory, COLT 2001 and 5th {E}uropean Conference on Computational Learning Theory, {EuroCOLT} 2001, Amsterdam, The Netherlands, July 2001, Proceedings VL - 2111 SP - 99 EP - 115 PB - Springer, Berlin N2 - In this paper we study a paradigm to generalize online classification algorithms for binary classification problems to multiclass problems. The particular hypotheses we investigate maintain one prototype vector per class. Given an input instance, a multiclass hypothesis computes a similarityscore between each prototype and the input instance and sets the predicted label to be the index of the prototype achieving the highest similarity. To design and analyze the learning algorithms in this... KW - machine-learning KW - mira AU - Crammer, Koby AU - Singer, Yoram PY - 2001/// UR - http://citeseer.ist.psu.edu/crammer03ultraconservative.html ER -

RIS BibTeX