Ordinal regression by a generalized force-based model

Francisco Fernández-Navarro, Annalisa Riccardi, Sante Carloni

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

This paper introduces a new instance-based algorithm for multiclass classification problems where the classes have a natural order. The proposed algorithm extends the state-of-the-art gravitational models by generalizing the scaling behavior of the class-pattern interaction force. Like the other gravitational models, the proposed algorithm classifies new patterns by comparing the magnitude of the force that each class exerts on a given pattern. To address ordinal problems, the algorithm assumes that, given a pattern, the forces associated to each class follow a unimodal distribution. For this reason, a weight matrix that allows to modify the metric in the attributes space and a vector of parameters that allows to modify the force law for each class have been introduced in the model definition. Furthermore, a probabilistic formulation of the error function allows the estimation of the model parameters using global and local optimization procedures toward minimization of the errors and penalization of the non unimodal outputs. One of the strengths of the model is its competitive grade of interpretability which is a requisite in most of real applications. The proposed algorithm is compared to other well-known ordinal regression algorithms on discretized regression datasets and real ordinal regression datasets. Experimental results demonstrate that the proposed algorithm can achieve competitive generalization performance and it is validated using nonparametric statistical tests.
LanguageEnglish
Pages844-857
Number of pages14
JournalIEEE Transactions on Cybernetics
Volume45
Issue number4
DOIs
Publication statusPublished - 24 Jul 2014
Externally publishedYes

Fingerprint

Statistical tests

Keywords

  • force
  • computational modeling
  • adaptation models
  • matrix algebra

Cite this

Fernández-Navarro, Francisco ; Riccardi, Annalisa ; Carloni, Sante. / Ordinal regression by a generalized force-based model. In: IEEE Transactions on Cybernetics. 2014 ; Vol. 45, No. 4. pp. 844-857.
@article{927d64a536b14cd38204ba0d1b5d14ff,
title = "Ordinal regression by a generalized force-based model",
abstract = "This paper introduces a new instance-based algorithm for multiclass classification problems where the classes have a natural order. The proposed algorithm extends the state-of-the-art gravitational models by generalizing the scaling behavior of the class-pattern interaction force. Like the other gravitational models, the proposed algorithm classifies new patterns by comparing the magnitude of the force that each class exerts on a given pattern. To address ordinal problems, the algorithm assumes that, given a pattern, the forces associated to each class follow a unimodal distribution. For this reason, a weight matrix that allows to modify the metric in the attributes space and a vector of parameters that allows to modify the force law for each class have been introduced in the model definition. Furthermore, a probabilistic formulation of the error function allows the estimation of the model parameters using global and local optimization procedures toward minimization of the errors and penalization of the non unimodal outputs. One of the strengths of the model is its competitive grade of interpretability which is a requisite in most of real applications. The proposed algorithm is compared to other well-known ordinal regression algorithms on discretized regression datasets and real ordinal regression datasets. Experimental results demonstrate that the proposed algorithm can achieve competitive generalization performance and it is validated using nonparametric statistical tests.",
keywords = "force, computational modeling, adaptation models, matrix algebra",
author = "Francisco Fern{\'a}ndez-Navarro and Annalisa Riccardi and Sante Carloni",
year = "2014",
month = "7",
day = "24",
doi = "10.1109/TCYB.2014.2337113",
language = "English",
volume = "45",
pages = "844--857",
journal = "IEEE Transactions on Cybernetics",
issn = "2168-2267",
number = "4",

}

Ordinal regression by a generalized force-based model. / Fernández-Navarro, Francisco; Riccardi, Annalisa; Carloni, Sante.

In: IEEE Transactions on Cybernetics, Vol. 45, No. 4, 24.07.2014, p. 844-857.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Ordinal regression by a generalized force-based model

AU - Fernández-Navarro, Francisco

AU - Riccardi, Annalisa

AU - Carloni, Sante

PY - 2014/7/24

Y1 - 2014/7/24

N2 - This paper introduces a new instance-based algorithm for multiclass classification problems where the classes have a natural order. The proposed algorithm extends the state-of-the-art gravitational models by generalizing the scaling behavior of the class-pattern interaction force. Like the other gravitational models, the proposed algorithm classifies new patterns by comparing the magnitude of the force that each class exerts on a given pattern. To address ordinal problems, the algorithm assumes that, given a pattern, the forces associated to each class follow a unimodal distribution. For this reason, a weight matrix that allows to modify the metric in the attributes space and a vector of parameters that allows to modify the force law for each class have been introduced in the model definition. Furthermore, a probabilistic formulation of the error function allows the estimation of the model parameters using global and local optimization procedures toward minimization of the errors and penalization of the non unimodal outputs. One of the strengths of the model is its competitive grade of interpretability which is a requisite in most of real applications. The proposed algorithm is compared to other well-known ordinal regression algorithms on discretized regression datasets and real ordinal regression datasets. Experimental results demonstrate that the proposed algorithm can achieve competitive generalization performance and it is validated using nonparametric statistical tests.

AB - This paper introduces a new instance-based algorithm for multiclass classification problems where the classes have a natural order. The proposed algorithm extends the state-of-the-art gravitational models by generalizing the scaling behavior of the class-pattern interaction force. Like the other gravitational models, the proposed algorithm classifies new patterns by comparing the magnitude of the force that each class exerts on a given pattern. To address ordinal problems, the algorithm assumes that, given a pattern, the forces associated to each class follow a unimodal distribution. For this reason, a weight matrix that allows to modify the metric in the attributes space and a vector of parameters that allows to modify the force law for each class have been introduced in the model definition. Furthermore, a probabilistic formulation of the error function allows the estimation of the model parameters using global and local optimization procedures toward minimization of the errors and penalization of the non unimodal outputs. One of the strengths of the model is its competitive grade of interpretability which is a requisite in most of real applications. The proposed algorithm is compared to other well-known ordinal regression algorithms on discretized regression datasets and real ordinal regression datasets. Experimental results demonstrate that the proposed algorithm can achieve competitive generalization performance and it is validated using nonparametric statistical tests.

KW - force

KW - computational modeling

KW - adaptation models

KW - matrix algebra

U2 - 10.1109/TCYB.2014.2337113

DO - 10.1109/TCYB.2014.2337113

M3 - Article

VL - 45

SP - 844

EP - 857

JO - IEEE Transactions on Cybernetics

T2 - IEEE Transactions on Cybernetics

JF - IEEE Transactions on Cybernetics

SN - 2168-2267

IS - 4

ER -