@article {10.3844/jcssp.2025.1688.1704,
article_type = {journal},
title = {Enhancing the CIC IoT Dataset 2023 for Improved Attack Detection through GANs Augmentation and Federated Learning},
author = {Alahmari, Shahad and Aleisa, Noura},
volume = {21},
number = {7},
year = {2025},
month = {Jul},
pages = {1688-1704},
doi = {10.3844/jcssp.2025.1688.1704},
url = {https://thescipub.com/abstract/jcssp.2025.1688.1704},
abstract = {The escalating frequency and sophistication of cyber-attacks on Internet of Things (IoT) devices present a pressing challenge to cybersecurity. With IoT device connections projected to exceed 42 billion by 2025, the vulnerability of these devices to cyber-attacks has never been more evident. This paper investigates the integration of Machine Learning (ML) and data augmentation, specifically Generative Adversarial Networks (GAN) and Federated Learning (FL), as innovative measures to fortify IoT security. The study aims to balance the CIC IoT Dataset 2023 using GAN-generated  synthetic data and to enhance ML model performance through FL, with eXtreme Gradient Boosting (XGBoost) as the FL framework's backbone. The utilization of GAN for data augmentation addresses the persistent challenge of data imbalances in datasets. The comparison between the FL and traditional approaches in IoT security analytics reveals distinctad vantages of FL, particularly in data privacy, scalability, and handling imbalanced data. While FL consistently delivers high accuracy, precision, recall, and F1-scores, the traditional approach varies more, often requiring additional data balancing and model tuning.},
journal = {Journal of Computer Science},
publisher = {Science Publications}
}