DSpace DSpace UPC
 Català   Castellano   English  

E-prints UPC >
Altres >
Enviament des de DRAC >

Empreu aquest identificador per citar o enllaçar aquest ítem: http://hdl.handle.net/2117/9799

Arxiu Descripció MidaFormat
JMLR2008.pdf483,45 kBAdobe PDFThumbnail
Veure/Obrir

Citació: Collins, M. [et al.]. Exponentiated gradient algorithms for conditional random fields and max-margin Markov networks. "Journal of Machine Learning Research", Agost 2008, vol. 9, p. 1775-1822.
Títol: Exponentiated gradient algorithms for conditional random fields and max-margin Markov networks
Autor: Collins, Michael; Globerson, Amir; Koo, Terry; Carreras Pérez, Xavier Veure Producció científica UPC; Bartlett, Peter
Data: ago-2008
Tipus de document: Article
Resum: Log-linear and maximum-margin models are two commonly-used methods in supervised machine learning, and are frequently used in structured prediction problems. Efficient learning of parameters in these models is therefore an important problem, and becomes a key factor when learning from very large data sets. This paper describes exponentiated gradient (EG) algorithms for training such models, where EG updates are applied to the convex dual of either the log-linear or max-margin objective function; the dual in both the log-linear and max-margin cases corresponds to minimizing a convex function with simplex constraints. We study both batch and online variants of the algorithm, and provide rates of convergence for both cases. In the max-margin case, O(1/ε) EG updates are required to reach a given accuracy ε in the dual; in contrast, for log-linear models only O(log(1/ε)) updates are required. For both the max-margin and log-linear cases, our bounds suggest that the online EG algorithm requires a factor of n less computation to reach a desired accuracy than the batch EG algorithm, where n is the number of training examples. Our experiments confirm that the online algorithms are much faster than the batch algorithms in practice. We describe how the EG updates factor in a convenient way for structured prediction problems, allowing the algorithms to be efficiently applied to problems such as sequence learning or natural language parsing. We perform extensive evaluation of the algorithms, comparing them to L-BFGS and stochastic gradient descent for log-linear models, and to SVM-Struct for max-margin models. The algorithms are applied to a multi-class problem as well as to a more complex large-scale parsing task. In all these settings, the EG algorithms presented here outperform the other methods.
ISSN: 1533-7928
URI: http://hdl.handle.net/2117/9799
Versió de l'editor: http://jmlr.csail.mit.edu/papers/v9/collins08a.html
Apareix a les col·leccions:GPLN - Grup de Processament del Llenguatge Natural. Articles de revista
Altres. Enviament des de DRAC
Comparteix:


Stats Mostra les estadístiques d'aquest ítem

SFX Query

Tots els drets reservats. Aquesta obra està protegida pels drets de propietat intel·lectual i industrial corresponents. Sense perjudici de les exempcions legals existents, queda prohibida la seva reproducció, distribució, comunicació pública o transformació sense l'autorització del titular dels drets.

Per a qualsevol ús que se'n vulgui fer no previst a la llei, dirigiu-vos a: sepi.bupc@upc.edu

 

Valid XHTML 1.0! Programari DSpace Copyright © 2002-2004 MIT and Hewlett-Packard Comentaris
Universitat Politècnica de Catalunya. Servei de Biblioteques, Publicacions i Arxius