Choquet integral based on copulas and its application in neural networks

i

This article has been updated. A Correction to this article was published on .

i

The Original Article was published on

i

The Original Article was published on

i

This article has been updated

i

This article was retracted on

i

This article has been updated

Abstract

Usually, convolutional neural networks (CNNs) reduce the features extracted from convolutional layers by adopting the maximum value or arithmetic average method, which is called the pooling process. However, these two pooling methods overlook the spatial dependencies between image features. The Choquet integral is a nonlinear integral, which addresses this limitation by incorporating weighted aggregation via a fuzzy measure, enabling it to effectively model interactions between variables. This capability is essential for processing data with complex dependency structures. Therefore, this paper proposes a novel Choquet integral pooling method for feature extraction in the pooling layer. By conducting experiments on datasets, we demonstrated the effectiveness of the proposed method and compared it with existing techniques. The experimental results indicated that for cases where image features are less prominent and exhibit spatial dependencies, the Choquet integral pooling outperforms traditional pooling methods and demonstrates greater robustness. To ensure the rationality and validity of applying the Choquet integral in CNNs, we conducted a more in-depth study of the Choquet integral based on a copula, that is, the C_C−integral, including its averaging, idempotence, translation invariance, and positive homogeneity. This research not only enriches the application fields of the Choquet integral but also provides new theoretical support and technical paths for the research of neural networks.

Keywords

fuzzy analysis / Choquet integral / C_C−integral / convolutional neural networks

CHANGE HISTORY

A Correction to this paper has been published:
https://doi.org

The Retraction Note was updated to reflect reception of statements of disagreement from all authors.

This article has been retracted.Please see the Retraction Notice for more detail:
https://doi.org

References

[1] G. Beliakov, H. Bustince, T. Calvo, A Practical Guide to Averaging Functions , Springer, Cham, Switzerland, 2016. https://doi.org/10.1007/978-3-319-24753-3

[2] E. Barrenechea, H. Bustince, J. Fernandez, D. Paternain, J. A. Sanz, Using the Choquet integral in the fuzzy reasoning method of fuzzy rule-based classification systems,Axioms, 2(2013), 208-223. https://doi.org/10.3390/axioms2020208

[3] D. Paternain, J. Fernandez, H. Bustince, R. Mesiar, G. Beliakov, Construction of image reduction operators using averaging aggregation functions,Fuzzy Sets Syst., 261(2015), 87-111. https://doi.org/10.1016/j.fss.2014.03.008

[4] L. M. Rodrigues, G. P. Dimuro, D. T. Franco, J. C. Fachinello, A system based on interval fuzzy approach to predict the appearance of pests in agriculture, in 2013 Joint IFSA World Congress and NAFIPS Annual Meeting (IFSA/NAFIPS), Edmonton, AB, Canada, 2003, 1262-1267. https://doi.org/10.1109/IFSA-NAFIPS.2013.6608583

[5] C. Campomanes-Alvarez, O. Ibanez, O. Cordon, C. Wilkinson, Hierarchical information fusion for decision making in craniofacial superimposition,Inf. Fusion, 39(2018), 25-40. https://doi.org/10.1016/j.inffus.2017.03.004

[6] Q. Zhang, L. Yang, Z. Chen, P. Li, A survey on deep learning for big data,Inf. Fusion, 42(2018), 146-157. https://doi.org/10.1016/j.inffus.2017.10.006

[7] G. Choquet, Theory of capacities, Ann. de Inst. Fourier, 5(1954), 131-295. https://doi.org/10.5802/AIF.53

[8] J. C. R. Alcantud, The capacity compliance problem: Refinements of time discounting to Choquet integrals with 2-additive fuzzy measures, Comput. Appl. Math., 44(2025), 424. https://doi.org/10.1007/s40314-025-03407-4

[9] M. Skublewska-Paszkowska, P. Karczmarek, P. Powroznik, E. Lukasik, J. Smolka, M. Dolecki, Aggregation of tennis multivariate time-series using the Choquet integral and its generalizations, in 2023 IEEE International Conference on Fuzzy Systems (FUZZ), Incheon, Korea, Republic of, 2023, 1-6. https://doi.org/10.1109/FUZZ52849.2023.10309746

[10] J. J. Huang, C. Y. Chen, Temporally adaptive hierarchical Choquet integrals: A measure-theoretic framework for dynamic non-additive integration in approximate reasoning,Fuzzy Sets Syst., 520(2025),109550. https://doi.org/10.1016/j.fss.2025.109550

[11] M. Sugeno, T. Murofushi, Pseudo-additive measure and integral, J. Math. Anal. Appl., 122(1987), 197-222. https://doi.org/10.1016/0022-247X(87)90354-4

[12] M. Grabisch, Fuzzy measure and integrals: Recent developments, in: D. Tamir, N. Rishe, A. Kandel (eds), inFifty Years of Fuzzy Logic and its Applications , Springer International, Cham, Switzerland, 2015, 9-22. https://doi.org/10.1007/978-3-319-19683-18

[13] R. Mesiar, Choquet-like integral, J. Math. Anal. Appl., 194(1995), 477-488. https://doi.org/10.1006/jmaa.1995.1312

[14] G. Lucca, J. A. Sanz, G. P. Dimuro, B. Bedregal, R. Mesiar, A. Koles árová, et al., Pre-aggregation functions: Construction and an application,IEEE Trans. Fuzzy Syst., 24(2016), 260-272. https://doi.org/10.1109/TFUZZ.2015.2453020

[15] G. Lucca, J. A. Sanz, G. P. Domuro, B. Bedregal, M. J. Asiain, M. Eikano, et al., Cc-Integral: Choquet-like copula-based aggregation functions and its application in fuzzy rule-based classification system, Knowl.-Based Syst., 119(2017), 32-43. https://doi.org/10.1016/j.knosys.2016.12.004

[16] G. Lucca, J. A. Sanz, G. P. Dimuro, B. Bedregal, H. Bustince, R. Mesiar, CF-integrals: A new family of pre-aggregation functions with application to fuzzy rule-based classification systems, Inf. Sci., 435(2018), 94-110. https://doi.org/10.1016/j.ins.2017.12.029

[17] P. G. Dimuro, J. Fernandez, B. Bedregal, R. Mesiar, J. A. Sanz, G. Lucca, et al., The state-of-art of the generalization of the Choquet integral: From aggregation and pre-aggregation to ordered directionally monotone function,Inf. Fusion, 57(2020), 27-43. https://doi.org/10.1016/j.inffus.2019.10.005

[18] A. Krizhevsky, I. Sutskever, G. E. Hinton, ImageNet classification with deep convolutional neural networks,Commun. ACM, 60(2017), 84-90. https://doi.org/10.1145/3065386

[19] R. Girshick, J. Donahue, T. Darrell, J. Malik, Rich feature hierarchies for accurate object detection and semantic segmentation, in 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, USA,2014, 580-587. https://doi.org/10.1109/CVPR.2014.81

[20] R. Collobert, J. Weston, A unified architecture for natural language processing: Deep neural networks with multitask learning, inICML ’08: Proceedings of the 25th International Conference on Machine Learning, 2008, 160-167. https://doi.org/10.1145/1390156.1390177

[21] M. Lin, Q. Chen, S. Yan, Network In Network, inInternational Conference on Learning Representations, 2014. Available from: https://openreview.net/forum?id=ylE6yojDR5yqX.

[22] D. Yu, H. Wang, P. Chen, Z. Wei, Mixed pooling for convolutional neural networks, in Rough Sets and Knowledge Technology, 8818(2014), 364-375. https://doi.org/10.1007/978-3-319-11740-934

[23] K. He, X. Zhang, S. Ren, J. Sun, Spatial pyramid pooling in deep convolutional networks for visual recognition,IEEE Trans. Pattern Anal. Mach. Intell., 37(2015), 1904-1916. https://doi.org/10.1109/TPAMI.2015.2389824

[24] B. Graham, Fractional max-pooling, preprint, arXiv:1412.6071.

[25] S. Sabour, N. Frosst, G. E. Hinton, Dynamic routing between capsules, inNIPS’17: Proceedings of the 31st International Conference on Neural Information Processing Systems , Red Hook, NY, USA, 2017, 3859-3869. Available from: https://dlnext.acm.org/doi/10.5555/3294996.3295142.

[26] C. Y. Lee, P. Gallagher, Z. Tu, Generalizing pooling functions in convolutional neural networks: Mixed, gated, and tree,IEEE Trans. Pattern Anal. Mach. Intell., 40(2018), 863-875. https://doi.org/10.1109/TPAMI.2017.2703082

[27] I. Rodriguez-Martinez, J. Lafuente, R. H. N. Santiago, G. P. Dimuro, F. P. Herrera, H. Bustince, Replacing pooling functions in Convolutional Neural Networks by linear combinations of increasing functions, Neural Networks, 152(2022), 380-393. https://doi.org/10.1016/j.neunet.2022.04.028

[28] I. Rodriguez-Martinez, T. da Cruz Asmus, G. P. Dimuro, F. Herrera, Z. Takáč, H. Bustince, Generalizing max pooling via (a,b)-grouping functions for convolutional neural networks,Inf. Fusion , 99(2023), 101893. https://doi.org/10.1016/j.inffus.2023.101893

[29] M. Ferrero-Jaurrieta, R. Paiva, A. Cruz, B. R. Callejas Bedregal, L. De Miguel, Z. Takáč, et al., Non-symmetric over-time pooling using pseudo-grouping functions for convolutional neural networks, Eng. Appl. Artif. Intell., 133(2024), 108470. https://doi.org/10.1016/j.engappai.2024.108470

[30] C. Dias, J. Bueno, E. Borges, G. Lucca, H. Santos, G. Dimuro, et al., Simulating the behaviour of Choquet-like (pre) aggregation gunctions for image resizing in the pooling layer of deep learning networks, inFuzzy Techniques: Theory and Applications, Springer, Cham, 2019, 224-236. https://doi.org/10.1007/978-3-030-21920-821

[31] C. A. Dias, J. C. S. Bueno, E. N. Borges, S. S. C. Botelho, G. P. Dimuro, G. Lucca, et al., Using the Choquet integral in the pooling layer in deep learning, inFuzzy Information Processing, Springer, Cham, 2018, 144-154. https://doi.org/10.1007/978-3-030-03023-014

[32] G. Beliakov, A. Pradera, T. Calvo, Aggregation Functions: A Guide for Practitioners , Springer Berlin, Heidelberg, 2007. https://doi.org/10.1007/978-3-540-73721-6

[33] E. P. Klement, R. Mesiar, E. Pap, Triangular Norms , Springer Dordrecht, 2000. https://doi.org/10.1007/978-94-015-9540-7

[34] G. Mayor, E. Trillas, On the representation of some aggregation functions, inProceedings of the 16th IEEE International Symposium on Multiple-Valued Logic (ISMVL), Los Alamitos, 1986, 110-114.

[35] C. Alsina, M. J. Frank, B. Schweizer, Associative Functions. Triangular Norms and Copulas , World Scientific, 2006, 1-360. https://doi.org/10.1142/9789812774200

[36] B. Schweizer, A. Sklar, Probabilistic Metric Spaces , North-Holland, New York, 1983.

[37] D. Gomez, J. T. Rodriguez, J. Montero, H. Bustince, E. Barrenechea, n-dimensional overlap functions,Fuzzy Sets Syst., 287(2016), 57-75. https://doi.org/10.1016/j.fss.2014.11.023

[38] M. Elkano, M. Galar, J. Sanz, H. Bustince, Fuzzy rule-based classification systems for multi-class problems using binary decomposition strategies: On the influence of n-dimensional overlap functions in the fuzzy reasoning method,Inf. Sci., 332(2016), 94-114. https://doi.org/10.1016/j.ins.2015.11.006

[39] M. Elkano, M. Galar, J. A. Sanz, A. Fernandez, E. Barrenechea, F. Herrera, et al., Enhancing multi-class classification in FARC-HD fuzzy classifier: on the synergy between n-dimensional overlap functions and decomposition strategies,IEEE Trans. Fuzzy Syst., 23(2015), 1562-1576. https://doi.org/10.1109/TFUZZ.2014.2370677

[40] H. Bustince, J. Fernandez, A. Kolasarova, R. Mesiar, Directional monotonicity of fusion,Eur. J. Oper. Res., 244(2015), 300-308. https://doi.org/10.1016/j.ejor.2015.01.018

[41] T. Wilkin, G. Beliakov, Weakly monotonic averaging functions,Int. J. Intell. Syst., 30(2015), 144-169. https://doi.org/10.1002/int.21692

[42] T. Murofushi, M. Sugeno, M. Machida, Non-monotonic fuzzy measures and the Choquet integral, Fuzzy Sets Syst., 64(1994), 73-84. https://doi.org/10.1016/0165-0114(94)90008-6

[43] H. Yang, L. Shang, Z. Gong, C_C−integral on interval-valued Sugeno probability measure and its application in multi-criteria decision-making rroblems, Axioms, 11(2022), 317. https://doi.org/10.3390/axioms11070317

[44] J. Qiao, B. Q. Hu, The distributive laws of fuzzy implications over overlap and grouping functions, Inf. Sci., 438(2018), 107-126. https://doi.org/10.1016/j.ins.2018.01.047

[45] H. Bustince, R. Mesiar, J. Fernandez, M. Galar, D. Paternain, A. Altalhi, et al. d-Choquet integrals: Choquet integrals based on dissimilarities,Fuzzy Sets Syst., 414(2021), 1-27. https://doi.org/10.1016/j.fss.2020.03.019