Evaluating Feature Combinations for Banknote Authentication Using Logistic Regression and Gaussian Naive Bayes
DOI:
https://doi.org/10.59992/IJCI.2026.v5n9p4الكلمات المفتاحية:
Banknote Authentication، Feature Combination، Logistic Regression، Gaussian Naive Bayes، Machine Learning، Classificationالملخص
The study examines the impact of feature combinations on banknote authentication using two lightweight and interpretable classifiers: Logistic Regression (LR) and Gaussian Naive Bayes (GNB). Experiments were performed on the Banknote Authentication dataset which has 1372 instances and four numerical features: Variance, Skewness, Kurtosis, and Entropy. Two, three, and four features were systematically evaluated in 11 that included combinations. Repeated stratified five-fold cross-validation with 10 repetitions was used to obtain robust estimates of performance. Evaluation measures used were accuracy, precision, recall, F1-Score and ROC-AUC. Logistic Regression performed the best with Variance, Skewness and Kurtosis with accuracy of 98.97% ± 0.56%, F1-Score of 98.85% and ROC-AUC of 99.98%. This same accuracy and F1-Score was achieved when all four features were used, suggesting that the performance gain of using Entropy was not noticeable for Logistic Regression. Gaussian Naives Bayes was able to get the highest accuracy with 87.22% ± 1.60% based on Variance, Skewness and Entropy. The results show that systematic feature-combination analysis is able to find compact representation and at the same time find that the best smaller feature subset varies according to the classifier.
المراجع
[1] F. Van Der Horst, J. Snell, and J. Theeuwes, “Enhancing banknote authentication by guiding attention to security features and manipulating prevalence expectancy,” Cogn. Res. Princ. Implic., 2021, doi: 10.1186/s41235-021-00341-x.
[2] N. G. Abadi, “Machine Learning-Based Authentication of Banknotes : A Comprehensive Analysis,” vol. 4, no. 1, pp. 22–30, 2024.
[3] V. Lohweg, “Banknote Authentication,” UCI Machine Learning Repository. [Online]. Available: https://archive.ics.uci.edu/dataset/267/banknote+authentication
[4] R. Nie, “Comparative Analysis Of Machine Learning Based Bank Note Authen- tication Through Variable Selection,” pp. 1–12, 2023.
[5] P. A. Singh, P. Z. Hasan, and P. S. Sharma, “Improving Banknote Authentication Accuracy through Logistic Regression and Data Preprocessing,” vol. 10, no. 1, 2022, doi: 10.15680/IJIRCCE.2022.1001044.
[6] K. Sn and K. Chitra, “Evaluating Machine Learning Algorithms for Fake Currency Detection,” vol. 2024, no. June, 2024.
[7] J. J. Jena, “A Comprehensive Exploration of Machine Learning Models for Banknote Authentication,” 2025 Int. Conf. Intell. Cloud Comput., pp. 1–6, 2025, doi: 10.1109/ICoICC64033.2025.11052182.
[8] O. V. Eşİdİr, G. Bak, and B. Karadağ, “Finansal Ve Medyatik Sahteciliğe Karşı Yapay Zekâ Çözümleri : Banknot Sahteciliği Tespitinde Makine Öğrenmesi Uygulamaları Artificial Intelligence Solutions Against Financial And Media Counterfeiting : Machine Learning Applications İn Banknote Counterfeit Detection,” vol. 17, no. 33, pp. 1066–1076, 2025.
[9] N. M. Z. Hamed and F. Al Azzo, “Develop a Robust System for Detecting Counterfeit Iraq Currencies Based on Deep Learning Techniques,” vol. 33, no. 2, 2024.
[10] J. K. C.-A. H. V. Lohweg, “Hidden in Plain Sight: Adversarial Attack on Wavelet-Based Banknote Authentication,” Res. Technol. Soc. Ind. Innov., 2024, doi: 10.1109/RTSI61910.2024.10761182.
[11] T. F. Document, “Identification of Counterfeit Currency Using Image Processing and Deep Learning,” IEEE 5th Int. Conf. Converg. Technol., 2019, doi: 10.1109/I2CT45611.2019.9033740.
[12] S. Sugahara and J. Cussens, “Learning Bayesian Network Classifiers to Minimize Class Variable Parameters,” vol. 27, pp. 1–41, 2026.
[13] A. Bhandare, S. Biradar, N. Lonari, and V. Pawar, “Online Subsidy Management System Using Machine Learning ( Algorithm- Logistic Regression , Random Forest , Decision Tree ),” vol. 3, no. 4, pp. 17–21, 2026, doi: 10.5281/zenodo.19899201.
[14] M. V. Anand, B. Kiranbala, S. R. Srividhya, C. Kavitha, and M. Younus, “Gaussian Na ¨ ıve Bayes Algorithm : A Reliable Technique Involved in the Assortment of the Segregation in Cancer,” vol. 2022, 2022, doi: 10.1155/2022/2436946.
