2015
Read, Jesse; Martino, Luca; Olmos, Pablo M; Luengo, David
Scalable Multi-Output Label Prediction: From Classifier Chains to Classifier Trellises Artículo de revista
En: Pattern Recognition, vol. 48, no 6, pp. 2096–2106, 2015.
Resumen | Enlaces | BibTeX | Etiquetas: Bayesian networks, classifer chains, Journal, Multi-label classification, multi-output prediction, structured inference
@article{Read2015b,
title = {Scalable Multi-Output Label Prediction: From Classifier Chains to Classifier Trellises},
author = {Jesse Read and Luca Martino and Pablo M Olmos and David Luengo},
url = {http://www.sciencedirect.com/science/article/pii/S0031320315000084},
year = {2015},
date = {2015-06-01},
journal = {Pattern Recognition},
volume = {48},
number = {6},
pages = {2096--2106},
abstract = {Multi-output inference tasks, such as multi-label classification, have become increasingly important in recent years. A popular method for multi-label classification is classifier chains, in which the predictions of individual classifiers are cascaded along a chain, thus taking into account inter-label dependencies and improving the overall performance. Several varieties of classifier chain methods have been introduced, and many of them perform very competitively across a wide range of benchmark datasets. However, scalability limitations become apparent on larger datasets when modeling a fully cascaded chain. In particular, the methods' strategies for discovering and modeling a good chain structure constitutes a mayor computational bottleneck. In this paper, we present the classifier trellis (CT) method for scalable multi-label classification. We compare CT with several recently proposed classifier chain methods to show that it occupies an important niche: it is highly competitive on standard multi-label problems, yet it can also scale up to thousands or even tens of thousands of labels.},
keywords = {Bayesian networks, classifer chains, Journal, Multi-label classification, multi-output prediction, structured inference},
pubstate = {published},
tppubtype = {article}
}
Multi-output inference tasks, such as multi-label classification, have become increasingly important in recent years. A popular method for multi-label classification is classifier chains, in which the predictions of individual classifiers are cascaded along a chain, thus taking into account inter-label dependencies and improving the overall performance. Several varieties of classifier chain methods have been introduced, and many of them perform very competitively across a wide range of benchmark datasets. However, scalability limitations become apparent on larger datasets when modeling a fully cascaded chain. In particular, the methods' strategies for discovering and modeling a good chain structure constitutes a mayor computational bottleneck. In this paper, we present the classifier trellis (CT) method for scalable multi-label classification. We compare CT with several recently proposed classifier chain methods to show that it occupies an important niche: it is highly competitive on standard multi-label problems, yet it can also scale up to thousands or even tens of thousands of labels.
Varando, Gherardo; Bielza, Concha; Larrañaga, Pedro
Decision boundary for discrete Bayesian network classifiers Artículo de revista
En: Journal of Machine Learning Research, 2015.
Resumen | Enlaces | BibTeX | Etiquetas: Bayesian networks, CASI CAM CM, CIG UPM, decision boundary, Journal, Lagrange basis, polynomial, supervised classi cation, threshold function
@article{Varando2015c,
title = {Decision boundary for discrete Bayesian network classifiers},
author = {Gherardo Varando and Concha Bielza and Pedro Larra\~{n}aga},
url = {http://cig.fi.upm.es/node/881 http://cig.fi.upm.es/articles/2015/Varando-2015-JMLR.pdf},
year = {2015},
date = {2015-01-01},
journal = {Journal of Machine Learning Research},
abstract = {Bayesian network classi ers are a powerful machine learning tool. In order to evaluate the expressive power of these models, we compute families of polynomials that sign-represent decision functions induced by Bayesian network classi ers. We prove that those families are linear combinations of products of Lagrange basis polynomials. In absence of V -structures in the predictor sub-graph, we are also able to prove that this family of polynomials does indeed characterize the speci c classi er considered. We then use this representation to bound the number of decision functions representable by Bayesian network classi ers with a given structure},
keywords = {Bayesian networks, CASI CAM CM, CIG UPM, decision boundary, Journal, Lagrange basis, polynomial, supervised classi cation, threshold function},
pubstate = {published},
tppubtype = {article}
}
Bayesian network classi ers are a powerful machine learning tool. In order to evaluate the expressive power of these models, we compute families of polynomials that sign-represent decision functions induced by Bayesian network classi ers. We prove that those families are linear combinations of products of Lagrange basis polynomials. In absence of V -structures in the predictor sub-graph, we are also able to prove that this family of polynomials does indeed characterize the speci c classi er considered. We then use this representation to bound the number of decision functions representable by Bayesian network classi ers with a given structure