Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{194158,
        author = {C.K. Joe Uzuegbu},
        title = {APPLICATION OF MACHINE LEARNING IN BIOMETRIC AUTHENTICATION FOR AN EDUCATIONAL SYSTEM},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {10},
        pages = {7721-7737},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=194158},
        abstract = {Educational institutions now rely on digital systems for teaching, attendance, examinations, records, and access control. Yet the way many of these systems confirm identity is still weak. Passwords can be shared. Identity cards can be borrowed. Manual checks take time and still miss impersonation. This study presents a simulation-based evaluation of machine-learning-enabled biometric authentication for educational use. The work focuses on three practical use cases: online examination sign-in, automated attendance, and campus access control. Four configurations were compared: facial verification using convolutional neural network (CNN) embeddings, fingerprint verification using support vector machine (SVM) classification on engineered descriptors, iris verification using random-forest classification on normalized texture summaries, and a privacy-aware multimodal fusion model that combines face and fingerprint scores. For each modality, 8,000 authentication transactions were modeled under standard conditions, followed by a degraded-condition stress test covering poor lighting, partial fingerprint capture, intermittent connectivity, and ordinary student movement. Under standard conditions, face, fingerprint, iris, and multimodal systems achieved accuracies of 97.0%, 95.5%, 96.2%, and 98.6%, respectively. Their equal error rates were 3.0%, 4.5%, 3.9%, and 1.7%, while mean authentication latencies were 184 ms, 126 ms, 163 ms, and 247 ms. Under degraded conditions, the multimodal configuration still achieved 96.3% accuracy, performed better than the unimodal alternatives, and maintained the lowest false-acceptance risk. Threshold analysis identified 0.50 as the most balanced default fusion threshold for educational deployment. A fairness-calibration step reduced the maximum modeled subgroup accuracy gap in the facial pipeline from 3.4 percentage points to 1.6 percentage points. The overall result is clear: biometrics can strengthen identity assurance in education, but the best outcomes come when technical performance is combined with privacy controls, audit logging, explainable review paths, and simple fallback options for failed captures.},
        keywords = {biometric authentication; machine learning; educational systems; online examinations; attendance automation; multimodal biometrics; privacy; fairness; face recognition; fingerprint recognition.},
        month = {March},
        }