Hybrid dropout for deep ordinal classification

Authors

Francisco Bérchez-Moreno

Francisco Moreno-Cano

David Guijo-Rubio

Víctor Manuel Vargas-Yun

Pedro Antonio Gutiérrez

César Hervás-Martínez

Conference Proceedings

Proceedings of the 18th international work-conference on artificial neural networks, IWANN 2025

ISSN 0302-9743

Vol. 16008

Pages 500–511

Abstract

This paper presents a new application of a hybrid dropout technique for Ordinal Classification (OC), based on a novel regularisation method. Unlike standard dropout, which ignores class ordering, this hybrid dropout integrates ordinal information by adjusting neurons dropout probabilities based on their correlation with target labels. We evaluate its effectiveness using a ResNet18 architecture over three new OC datasets and compare it with the standard dropout approach and with an architecture with no dropout. Results show that the hybrid dropout consistently achieves the best performance across multiple well-known metrics (1-off, QWK, MAE, AMAE, and RPS), while also reducing prediction variability. Statistical analysis using the Wilcoxon signed-rank test confirms its robustness, obtaining 21 significant wins out of 30 comparisons, with no losses. These results highlight the importance of designing regularisation strategies that consider the problems ordinal structure, demonstrating that hybrid dropout effectively enhances generalisation and predictive accuracy.

BibTex Citation

@inproceedings{Berchez2025Hybrid,
	author = {B{\' e}rchez-Moreno, Francisco and Moreno-Cano, Francisco and Guijo-Rubio, David and Vargas-Yun, V{\' i}ctor Manuel and Guti{\' e}rrez, Pedro Antonio and Herv{\' a}s-Mart{\' i}nez, C{\' e}sar},
	booktitle = {Proceedings of the 18th international work-conference on artificial neural networks, {IWANN} 2025},
	doi = {10.1007/978-3-032-02725-2_39},
	year = {2025},
	pages = {500--511},
	organization = {Springer},
	title = {Hybrid dropout for deep ordinal classification},
	url = {doi.org/10.1007/978-3-032-02725-2_39},
	volume = {16008},
}

BibTex Unicode Citation

@inproceedings{Berchez2025Hybrid,
	author = {Bérchez-Moreno, Francisco and Moreno-Cano, Francisco and Guijo-Rubio, David and Vargas-Yun, Víctor Manuel and Gutiérrez, Pedro Antonio and Hervás-Martínez, César},
	booktitle = {Proceedings of the 18th international work-conference on artificial neural networks, {IWANN} 2025},
	doi = {10.1007/978-3-032-02725-2_39},
	year = {2025},
	pages = {500--511},
	organization = {Springer},
	title = {Hybrid dropout for deep ordinal classification},
	url = {doi.org/10.1007/978-3-032-02725-2_39},
	volume = {16008},
}

APA Citation

Bérchez-Moreno, F., Moreno-Cano, F., Guijo-Rubio, D., Vargas-Yun, V. M., Gutiérrez, P. A., & Hervás-Martínez, C. (2025). Hybrid dropout for deep ordinal classification. Proceedings of the 18th International Work-Conference on Artificial Neural Networks, IWANN 2025, 16008, 500–511. https://doi.org/10.1007/978-3-032-02725-2_39

RIS Citation

TY  - CONF
AU  - Bérchez-Moreno, Francisco
AU  - Moreno-Cano, Francisco
AU  - Guijo-Rubio, David
AU  - Vargas-Yun, Víctor Manuel
AU  - Gutiérrez, Pedro Antonio
AU  - Hervás-Martínez, César
C3  - Proceedings of the 18th international work-conference on artificial ne
ural networks, IWANN 2025
DA  - 2025///
C2  - 2025
DO  - 10.1007/978-3-032-02725-2_39
ID  - temp_id_707023696142
PB  - Springer
SP  - 500-511
TI  - Hybrid dropout for deep ordinal classification
UR  - doi.org/10.1007/978-3-032-02725-2_39
VL  - 16008
ER  -

CV Citation

F. Bérchez-Moreno (CA), F. Moreno-Cano, D. Guijo-Rubio, V.M. Vargas-Yun, P.A. Gutiérrez, C. Hervás-Martínez (4/6). "Hybrid dropout for deep ordinal classification". Proceedings of the 18th international work-conference on artificial neural networks, IWANN 2025, pp. 500–511, 2025.