Ordinal Regression Methods: Survey and Experimental Study

Gutiérrez, Pedro Antonio; Pérez-Ortiz, María; Sánchez‐Monedero, Javier; Fernández-Navarro, Francisco; Hervás-Martı́nez, César

doi:10.1109/tkde.2015.2457911

Cited by 351 publications

(273 citation statements)

References 103 publications

Supporting

Mentioning

269

Contrasting

Unclassified

Order By: Relevance

“…Other methods in the literature have not been included in the experiments because they are based on other classification paradigms (such as Gaussian processes [17]) or focused on other slightly different settings such as the transductive one [19,20] or ordinal 390 clustering [21]. Note that the ordinal version of Gaussian processes has been already compared to the ordinal version of kernel discriminant analysis, resulting in worst performance for real ordinal datasets with a much higher computational cost [11]. Finally, we include four different versions of our proposals (S-DL, CES-DL, ES-DL and KES-DL), where the main differences revolve around the 395 space in which the neighbourhood information is computed for the inclusion of unlabelled data and the use of a kernel learning strategy to optimise the kernel parameters.…”

Section: Methodologies Testedmentioning

confidence: 99%

“…Semi-supervised learning has being mainly studied for binary classification [5,6] and regression [2], al- 20 though recently the main focus has shifted to multi-class problems [7,8,9] (and even multi-dimensional ones [10]). This paper tackles the use of unlabelled data in the context of ordinal classification [11], a learning paradigm which shares properties of both classification and regression.…”

Section: Introductionmentioning

confidence: 99%

“…In contrast to multinomial classification, there exists some ordering among the elements of Y (the labelling space) and both standard classifiers and the zero-one loss function do not capture and reflect this ordering appropriately [11] (leading to worse models in terms of errors in the ordinal 30 scale). Concerning regression, Y is a non-metric space.…”

Section: Introductionmentioning

confidence: 99%

“…40 Several issues must be highlighted when developing new ordinal classifiers in order to exploit the presence of this order among categories. Firstly, this implicit data structure should be learned by the classifier in order to minimise the different ordinal classification errors [11], and, secondly, different evaluation measures or metrics should be developed in this context. The most popular approach for with respect to other ordinal classification methods, i.e.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Semi-supervised learning for ordinal Kernel Discriminant Analysis

et al. 2016

View full text Add to dashboard Cite

Ordinal classification considers those classification problems where the labels of the variable to predict follow a given order. Naturally, labelled data is scarce or difficult to obtain in this type of problems because, in many cases, ordinal labels are given by an user or expert (e.g. in recommendation systems). Firstly, this paper develops a new strategy for ordinal classification where both labelled and unlabelled data are used in the model construction step (a scheme which is referred to as semi-supervised learning). More specifically, the ordinal version of kernel discriminant learning is extended for this setting considering the neighbourhood information of unlabelled data, which is proposed to be computed in the feature space induced by the kernel function. Secondly, a new method for semi-supervised kernel learning is devised in the context of ordinal classification, which is combined with our developed classification strategy to optimise the kernel parameters. The experiments conducted compare 6 different approaches for semi-supervised learning in the context of ordinal classification in a battery of 30 datasets, showing 1) the good synergy of the ordinal version of discriminant analysis and the use of unlabelled data and 2) the advantage of computing distances in the feature space induced by the kernel function.

show abstract

Section: Methodologies Testedmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Semi-supervised learning for ordinal Kernel Discriminant Analysis

et al. 2016

View full text Add to dashboard Cite

show abstract

“…Examples are found in such areas as operations research [3], genetics [4], environmental science [5], meteorology [6], psychology [7], and many others. One can find very large-scale MR problems, e.g., in machine learning [8][9][10] and computer simulations [11].…”

Section: Introductionmentioning

confidence: 99%

A Dual Active-Set Algorithm for Regularized Monotonic Regression

Burdakov

Sysoev

2017

J Optim Theory Appl

View full text Add to dashboard Cite

Monotonic (isotonic) regression is a powerful tool used for solving a wide range of important applied problems. One of its features, which poses a limitation on its use in some areas, is that it produces a piecewise constant fitted response. For smoothing the fitted response, we introduce a regularization term in the monotonic regression, formulated as a least distance problem with monotonicity constraints. The resulting smoothed monotonic regression is a convex quadratic optimization problem. We focus on the case, where the set of observations is completely (linearly) ordered. Our smoothed pool-adjacent-violators algorithm is designed for solving the regularized problem. It belongs to the class of dual active-set algorithms. We prove that it converges to the optimal solution in a finite number of iterations that does not exceed the problem size. One of its advantages is that the active set is progressively enlarging by including one or, typically, more constraints per iteration. This resulted in solving large-scale test problems in a few iterations, whereas the size of that problems was prohibitively too large for the conventional quadratic optimization solvers. Although the complexity of our algorithm grows quadratically with the problem size, we found its running time to grow almost linearly in our computational experiments.

show abstract

Predicting pathogenicity of missense variants with weakly supervised regression

et al. 2019

View full text Add to dashboard Cite

Quickly growing genetic variation data of unknown clinical significance demand computational methods that can reliably predict clinical phenotypes and deeply unravel molecular mechanisms. On the platform enabled by the Critical Assessment of Genome Interpretation (CAGI), we develop a novel "weakly supervised" regression (WSR) model that not only predicts precise clinical significance (probability of pathogenicity) from inexact training annotations (class of pathogenicity) but also infers underlying molecular mechanisms in a variant-specific manner. Compared to multiclass logistic regression, a representative multiclass classifier, our kernelized WSR improves the performance for the ENIGMA Challenge set from 0.72 to 0.97 in binary area under the receiver operating characteristic curve (AUC) and from 0.64 to 0.80 in ordinal multiclass AUC. WSR model interpretation and protein structural interpretation reach consensus in corroborating the most probable molecular mechanisms by which some pathogenic BRCA1 variants confer clinical significance, namely metal-binding disruption for p.C44F and p.C47Y, protein-binding disruption for p.M18T, and structure destabilization for p.S1715N.

show abstract

Ordinal Regression Methods: Survey and Experimental Study

Cited by 351 publications

References 103 publications

Semi-supervised learning for ordinal Kernel Discriminant Analysis

Semi-supervised learning for ordinal Kernel Discriminant Analysis

A Dual Active-Set Algorithm for Regularized Monotonic Regression

Predicting pathogenicity of missense variants with weakly supervised regression

Contact Info

Product

Resources

About