Article Per Year (5 Year)
p-Index From 2020 - 2025
0.23
P-Index
This Author published in this journals
All Journal Journal of Computer Science and Research
Marbun, Ramlan
Unknown Affiliation
Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Economics, Econometrics & Finance Library & Information Science

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search
Journal : Journal of Computer Science and Research

 1 
Smarter School Labs: Fast and Accurate Anomaly Detection Using Lightweight CNN Technology Marbun, Ramlan; Iqbal, Muhammad
Journal of Computer Science and Research (JoCoSiR) Vol. 3 No. 1 (2025): Jan: Computer Science
Publisher : Asosiasi Perguruan Tinggi Informatika dan Ilmu Komputer (APTIKOM) Provinsi Sumatera Utara

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

This study proposes a lightweight convolutional neural network (CNN) model for anomaly detection in school computer laboratories, aiming to enhance operational reliability and cybersecurity awareness. Real-time event logs were collected from 20 computers (PC01–PC20) at Santo Nicholas School with slight variations in CPU, RAM, and network behavior to simulate real-world heterogeneity. After preprocessing and normalization, the merged dataset contained over 10,000 log entries labeled as normal or anomalous. The proposed lightweight CNN achieved 92.23% F1-score, 91.80% accuracy, and a false positive rate (FPR) of 18.47%, demonstrating a balance between detection precision and computational efficiency. Comparative evaluation shows that this architecture performs competitively while requiring fewer parameters and lower inference latency than conventional CNNs. The results highlight the suitability of the proposed model for deployment in low-resource educational environments, supporting early anomaly detection and preventive maintenance. Future research will explore cross-domain generalization and lightweight deployment through edge-AI integration.
Co-Authors Muhammad Iqbal