Finite Difference Gradient Estimation for Logistic Regression: Application to SARS-CoV-2 Genomic Classification

Halimatus Sa`diyah, Mohammad Jamhuri, Fachrur Rozi

Abstract


Gradient-based optimization conventionally relies on closed-form analytical derivatives, which are unavailable for many modern or non-differentiable model architectures. This paper proposes \emph{forward finite difference} (FFD) gradient estimation as a derivative-free training alternative and validates it rigorously on logistic regression — a model whose known analytical gradient enables direct verification of the numerical approximation. A formal $\mathcal{O}(h)$ error bound is proved and confirmed empirically, showing that gradient direction is faithfully preserved across a wide range of step sizes. The framework is applied to binary genomic classification of SARS-CoV-2 versus non-SARS-CoV-2 coronaviruses using normalized $4$-mer frequency profiles. The FFD optimizer achieves classification performance statistically equivalent to analytical gradient descent ($F_1 \geq 0.999$), while an ablation study demonstrates that nucleotide composition — not sequence length — drives
discrimination. External validation on unseen coronavirus lineages reveals strong generalization except for MERS-CoV, whose phylogenetic proximity to SARS-CoV-2 produces overlapping $k$-mer signatures. These results establish FFD logistic regression as a principled derivative-free baseline and motivate its extension to architectures where analytical gradients are intractable.

Keywords


Finite Difference Method; Derivative-Free Optimization; Logistic Regression; k-mer Frequency; Genomic Classification; SARS-CoV-2

Full Text:

PDF

References


[1] B. Bakhshandeh, Z. Jahanafrooz, A. Abbasi, M. B. Goli, M. Sadeghi, M. S. Mottaqi, and M. Zamani. Mutations in SARS-CoV-2; consequences in structure, function, and pathogenicity of the virus. Microbial Pathogenesis, 154: 104831, 2021. doi: 10.1016/j.micpath.2021.104831.

[2] H. Yang and Z. Rao. Structural biology of SARS-CoV-2 and implications for therapeutic development. Nature Reviews Microbiology, 19(11):685–700, 2021. doi: 10.1038/s41579-021-00630-8.

[3] M. A. Awal, M. Masud, M. S. Hossain, A. A. M. Bulbul, S. M. H. Mahmud, and A. K. Bairagi. A novel Bayesian optimization-based machine learning framework for COVID-19 detection from inpatient facility data. IEEE Access, 9:10263–10281, 2021. doi: 10.1109/ACCESS.2021.3050852.

[4] P. E. Fleitas, J. A. Paz, M. I. Simoy, C. Vargas, R. O. Cimino, A. J. Krolewiecki, and J. P. Aparicio. Clinical diagnosis of COVID-19: A multivariate logistic regression analysis. Germs, 11(2):221–237, 2021. doi: 10.18683/germs.2021.1259.

[5] J. Rudar, P. Kruczkiewicz, O. Vernygora, et al. Sequence signatures within SARS-CoV-2 genome. Microbiology Spectrum, 12(4):e03584–23, 2024. doi: 10.1128/spectrum.03584-23.

[6] S. Aminah, G. Ardaneswari, M. K. Awang, M. A. Yusaputra, and D. P. Sari. Protein sequence-based COVID-19 detection: A comparative study of machine learning classification methods. Journal of Electrical and Computer Engineering, 2024:8683822, 2024. doi: 10.1155/2024/8683822.

[7] A. Ghosh, M. Jaenada, and L. Pardo. Classification of COVID-19 patients using robust logistic regression. Journal of Statistical Theory and Practice, 16(4):57, 2022. doi: 10.1007/s42519-022-00295-3.

[8] S. Kaur, S. Gupta, D. Gupta, S. Juneja, A. Nauman, M. Khan, I. Husain, A. Islam, and S. Mallik. High-accuracy lung disease classification via logistic regression and advanced feature extraction techniques. Egyptian Informatics Journal, 29:100596, 2025. doi: 10.1016/j.eij.2024.100596.

[9] A. Ramírez Varela, S. Moreno López, S. Contreras-Arrieta, et al. Prediction of SARS-CoV-2 infection with a symptoms-based model. Preventive Medicine Reports, 27:101798, 2022. doi: 10.1016/j.pmedr.2022.101798.

[10] B. A. Sokhansanj and G. L. Rosen. Predicting COVID-19 severity from spike protein sequence. Computers in Biology and Medicine, 149:105969, 2022. doi: 10.1016/j.compbiomed.2022.105969.

[11] M. R. Romadhon and F. Kurniawan. A comparison of Naive Bayes, logistic regression and KNN for classifying breast cancer. In 2021 International Conference on Electrical and Information Technology (EIConCIT), pages 41–44. IEEE, 2021. doi: 10.1109/EIConCIT50028.2021.9431845.

[12] Mohammed J. Zaki and Wagner Meira. Data Mining and Machine Learning: Fundamental Concepts and Algorithms. Cambridge University Press, Cambridge, 2nd edition, 2020. doi: 10.1017/9781108564175.

[13] Richard L. Burden and J. Douglas Faires. Numerical Analysis. Brooks/Cole, Cengage Learning, Boston, 9th edition, 2011. ISBN 978-0-538-73351-9. URL https://www.worldcat.org/isbn/9780538733519.

[14] Jorge Nocedal and Stephen J. Wright. Numerical Optimization. Springer, New York, 2nd edition, 2006. doi: 10.1007/978-0-387-40065-5.

[15] Laith Alzubaidi, Jian Zhang, Ahmed J. Humaidi, Ayad Al-Dujaili, Ye Duan, Omran Al-Shamma, Javier Santamaría, Mohammed A. Fadhel, Muthana Al-Amidie, and Laith Farhan. Review of deep learning: concepts, CNN architectures, challenges, applications, future directions. Journal of Big Data, 8(1):53, 2021. doi: 10.1186/s40537-021-00444-8.

[16] David E. Rumelhart, Geoffrey E. Hinton, and Ronald J. Williams. Learning representations by back-propagating errors. Nature, 323:533–536, 1986. doi: 10.1038/323533a0.

[17] O. P. Singh, M. Vallejo, I. M. El-Badawy, et al. Classification of SARS-CoV-2 using machine learning. Computers in Biology and Medicine, 136:104650, 2021. doi: 10.1016/j.compbiomed.2021.104650.

[18] C. Moeckel, M. Mareboina, M. A. Konnaris, C. S. Y. Chan, I. Mouratidis, A. Montgomery, N. Chantzi, G. A. Pavlopoulos, and I. Georgakopoulos-Soares. A survey of k-mer methods and applications in bioinformatics. Computational and Structural Biotechnology Journal, 23: 2289–2303, 2024. doi: 10.1016/j.csbj.2024.05.025.




DOI: https://doi.org/10.18860/jrmm.v5i5.43509

Refbacks

  • There are currently no refbacks.