Machine Learning for Handwritten Digit Recognition: Enhancing J2 Criterion Feature Compression with Multi-Descriptor Fusion

Abstract

This paper extends a prior study on J2 criterion feature compression for handwritten digit recognition by investigating three enhancement directions identified as future work: spatial pyramid extension of Local Binary Patterns, the effect of training set size on compression quality, and multi-descriptor feature fusion incorporating Scale Invariant Feature Transform features. The baseline pipeline extracts hand-crafted features, compresses them into a nine-dimensional discriminant subspace using a from-scratch implementation of the J2 criterion, and classifies the result using a Support Vector Machine and Linear Discriminant Analysis. Spatial Pyramid Local Binary Pattern encoding raises the accuracy of texture-only recognition from 56.25 percent to 92.70 percent by preserving the spatial arrangement of local texture that a global histogram discards. A data scaling study across 5,000 to 60,000 training samples shows that recognition accuracy improves from 94.15 percent to 96.75 percent while the J2 criterion value remains approximately stable, confirming that the compression quality is largely data-volume independent for a fixed number of classes. A triple-descriptor fusion combining Spatial Pyramid Local Binary Pattern, Histogram of Oriented Gradients, and Scale Invariant Feature Transform Bag of Words features achieve the highest accuracy of 97.30 percent with the largest J2 value of 38.17, demonstrating that the three descriptors carry complementary discriminative information. The results validate the J2 compression framework as an effective and scalable dimensionality reduction approach across all tested feature configurations.

REFERENCES

[1] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,” Proceedings of the IEEE, vol. 86, no. 11, pp. 2278–2324, 1998.

[2] R. Plamondon and S. N. Srihari, “Online and offline handwriting recognition: A comprehensive survey,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 22, no. 1, pp. 63–84, 2000.

[3] A. Baldominos, Y. Saez, and P. Isasi, “A survey of handwritten character recognition with MNIST and EMNIST,” Applied Sciences, vol. 9, no. 15, p. 3169, 2019.

[4] C. Rudin, “Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead,” Nature Machine Intelligence, vol. 1, no. 5, pp. 206–215, 2019.

[5] S. T. Hasan, "Natural language processing for employee relations case management: Analyzing workplace complaints, grievances, and investigation narratives in healthcare organizations," Journal of Computational Systems & Engineering Insights, vol. 1, no. 1, pp. 11-24, 2023.

 [6] U. Iqbal, "AI-powered supplier risk intelligence: Predicting financial and geopolitical supply chain disruptions in U.S. critical industries," *Journal of Engineering and Computational Intelligence Review*, vol. 3, no. 2, pp. 173–193, 2025.

[7] P. N. Belhumeur, J. P. Hespanha, and D. J. Kriegman, “Eigenfaces vs. Fisherfaces: Recognition using class specific linear projection,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 19, no. 7, pp. 711–720, 1997.

[8] R. Parveen, "An explainable AI framework for ethical fraud prevention in U.S. federal welfare programs," *International Journal of Applied Mathematics*, vol. 38, no. 11S, pp. 1425–1435, 2025.

[9] S. T. Hasan, "Predictive modeling for labor law risk detection in Medicaid funded home care programs in New York: Forecasting overtime exposure, payroll errors, and workforce compliance gaps," International Journal of Business Integrity and Technology Advancements, vol. 1, no. 1, pp. 01-10, 2025.

[10] S. Akter, "Exploring the role of artificial intelligence in food waste reduction: Implications for public health, social behavior, and sustainable communities," Journal of Computational Systems & Engineering Insights, vol. 1, no. 1, pp. 25-36, 2023.

[11] M. M. Islam, "AI-driven fraud detection and prevention using human behavior analysis to enhance U.S. social and financial security," *International Journal of Applied Mathematics*, vol. 38, no. 8S, pp. 861–871, 2025.

[12] U. Iqbal, "AI-driven predictive maintenance for U.S. smart manufacturing: Deep learning models for equipment failure prediction and operational resilience," *Journal of Engineering and Computational Intelligence Review*, vol. 3, no. 1, pp. 114–138, 2025.

[13] G. Csurka, C. Dance, L. Fan, J. Willamowski, and C. Bray, “Visual categorisation with bags of keypoints,” in Proc. ECCV Workshop Statistical Learning in Computer Vision, 2004, pp. 1–22.

[14] I. Kate, "Marketing innovation in the US e-commerce sector: Strategies for success in the evolving digital environment," International Journal of Business Integrity and Technology Advancements, vol. 1, no. 2, pp. 01-07, 2025.

[15] A. Gupta et al., "A Facile approach to recycling used facemasks for high-performance energy-storage devices," in Specialty Polymers, CRC Press, 2023, pp. 427-443.

[16] S. M. H. Shah, F. Amin, and A. Khan, "Cyber-resilient mobile edge computing: A deep neural approach for secure and efficient task offloading," *The Asian Bulletin of Big Data Management*, vol. 5, no. 1, pp. 200–215, 2025.

[17] I. A. Badhan, M. N. Hasnain, and M. H. Rahman, "Enhancing operational efficiency: A comprehensive analysis of machine learning integration in industrial automation," *Journal of Business Insight and Innovation*, vol. 1, no. 2, pp. 61–77, 2022.

[18] U. Imtiaz, B. Ahmad, M. H. Sajid, Q. Abbas, M. A. Qureshi, S. Rasheed, and A. Khan, "An integrated machine learning framework for structural health monitoring of bridges: A case study on Soan Bridge," *The Asian Bulletin of Big Data Management*, vol. 5, no. 2, pp. 194–207, 2025.

[19] F. Pedregosa et al., “Scikit-learn: Machine learning in Python,” J. Mach. Learn. Res., vol. 12, pp. 2825–2830, 2011.

[20] G. Bradski, “The OpenCV Library,” Dr. Dobb’s Journal of Software Tools, vol. 25, no. 11, pp. 120–125, 2000.

[21] D. Impedovo and G. Pirlo, “Machine learning for handwriting recognition,” Entropy, vol. 20, no. 11, p. 821, 2018.

[22] F. Amin, M. A. But, I. Amin, and A. Khan, "The tokenized business marketplace: A blockchain and AI-powered framework for democratizing business ownership and investment," *International Journal of Business and Management Sciences*, vol. 5, no. 4, pp. 318–328, 2024.

[23] U. Imtiaz, "Investigating the impact of phishing attacks on organizational cybersecurity posture," *Spectrum of Engineering Sciences*, pp. 1758–1780, 2025.

[24] I. A. Badhan, M. N. Hasnain, and M. H. Rahman, "Advancing operational efficiency: An in-depth study of machine learning applications in industrial automation," *Policy Research Journal*, vol. 1, no. 2, pp. 21–41, 2023.

[25] S. Bhardwaj et al., "Bimetallic Co–Fe sulfide and phosphide as efficient electrode materials for overall water splitting and supercapacitor," Discover Nano, vol. 18, no. 1, p. 59, 2023.

[26] X. Chen, L. Jin, Y. Zhu, C. Luo, and T. Wang, “Text recognition in the wild: A survey,” ACM Computing Surveys, vol. 54, no. 2, pp. 1–35, 2021.

[27] M. A. Hearst, S. T. Dumais, E. Osuna, J. Platt, and B. Scholkopf, “Support vector machines,” IEEE Intelligent Systems, vol. 13, no. 4, pp. 18–28, 1998.