Improving Online Exam Verification with Class-Weighted and Augmented CNN Models
Jurnal Teknologi Informasi dan Terapan · 17.10.2025 · 11 (2) · 91-98
DOI: 10.25047/jtit.v11i2.435 ↗Kurz gefasst
Fachartikel in der Zeitschrift Jurnal Teknologi Informasi dan Terapan, online erschienen am 17.10.2025. Verfasst von Ilham Fanani und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
The COVID-19 pandemic has shifted interactions to virtual platforms, significantly impacting education, particularly online exams. However, these online exams have vulnerabilities, including exam jockeys. This study proposes a face classification model using a Convolutional Neural Network (CNN) to verify online exam takers. The model uses preprocessing techniques, i.e. normalization, data augmentation, and class weighting, to balance data and enhance generalization utilizing TensorFlow. The results show an overall accuracy of 85%, with a precision of 86.34%, a recall of 84.24%, an F1-score of 85.28% for legal takers, and a precision of 83.65%, recall of 85.81%, and an F1-score of 84.71% for illegal takers. These results indicate the model's balanced performance between legal and illegal classes. By integrating CNN with tailored preprocessing and training strategies, this study addresses gaps in existing authentication methods, offering a robust approach to online exam verification. The proposed model shows a chance for practical applications. However, further optimization through larger datasets and advanced augmentation techniques is recommended to improve its accuracy and adaptability to diverse real-world contexts
Zitieren
APA
Fanani, I., & Rianto, R. (2025). Improving Online Exam Verification with Class-Weighted and Augmented CNN Models. Jurnal Teknologi Informasi dan Terapan, 11(2), 91-98. https://doi.org/10.25047/jtit.v11i2.435
Vancouver
Fanani I, Rianto R. Improving Online Exam Verification with Class-Weighted and Augmented CNN Models. Jurnal Teknologi Informasi dan Terapan. 2025;11(2):91-98. doi:10.25047/jtit.v11i2.435
BibTeX
@article{Fanani2025,
author = {Fanani, Ilham and Rianto, Rianto},
title = {Improving Online Exam Verification with Class-Weighted and Augmented CNN Models},
journal = {Jurnal Teknologi Informasi dan Terapan},
year = {2025},
volume = {11},
number = {2},
pages = {91--98},
doi = {10.25047/jtit.v11i2.435}
}
EUVIMED