TY - GEN
T1 - ENIDSGAN
T2 - 2025 IEEE Annual Congress on Artificial Intelligence of Things, AIoT 2025
AU - Hartono, Darren
AU - Fang, Dongfeng
AU - Xu, Shengjie
N1 - Publisher Copyright: © 2025 IEEE.
PY - 2025
Y1 - 2025
N2 - Network Intrusion Detection Systems (NIDS) are essential for identifying and mitigating malicious activities targeting IoT devices. Traditionally, these systems have relied on signature-based and anomaly-based detection techniques. However, the increasing complexity and adaptability of cyber threats have driven the adoption of Machine Learning (ML) approaches in modern NIDS, significantly improving their ability to detect a wider range of attack vectors. Despite these advancements, ML-based NIDS remain vulnerable to adversarial examples, inputs designed to mislead models and trigger incorrect classifications. Adversarial examples now pose a critical threat to the robustness of ML-based cybersecurity systems. This work explores using Generative Adversarial Networks (GANs) to efficiently create adversarial examples that can evade detection while remaining valid on IoT networks. This work adapts, enhances, and evaluates a proposed GAN-based adversarial example generation algorithm to improve its effectiveness. Experimental tests demonstrate that the proposed algorithm outperforms existing methods, achieving a better attack success rate with reduced computational overhead.
AB - Network Intrusion Detection Systems (NIDS) are essential for identifying and mitigating malicious activities targeting IoT devices. Traditionally, these systems have relied on signature-based and anomaly-based detection techniques. However, the increasing complexity and adaptability of cyber threats have driven the adoption of Machine Learning (ML) approaches in modern NIDS, significantly improving their ability to detect a wider range of attack vectors. Despite these advancements, ML-based NIDS remain vulnerable to adversarial examples, inputs designed to mislead models and trigger incorrect classifications. Adversarial examples now pose a critical threat to the robustness of ML-based cybersecurity systems. This work explores using Generative Adversarial Networks (GANs) to efficiently create adversarial examples that can evade detection while remaining valid on IoT networks. This work adapts, enhances, and evaluates a proposed GAN-based adversarial example generation algorithm to improve its effectiveness. Experimental tests demonstrate that the proposed algorithm outperforms existing methods, achieving a better attack success rate with reduced computational overhead.
KW - Adversary Examples
KW - Generative Adversarial Networks
KW - Machine Learning
KW - Network Intrusion Detection
KW - Security
UR - https://www.scopus.com/pages/publications/105035765670
UR - https://www.scopus.com/pages/publications/105035765670#tab=citedBy
U2 - 10.1109/AIoT66900.2025.00080
DO - 10.1109/AIoT66900.2025.00080
M3 - Conference contribution
T3 - Proceedings - 2025 IEEE Annual Congress on Artificial Intelligence of Things, AIoT 2025
SP - 528
EP - 533
BT - Proceedings - 2025 IEEE Annual Congress on Artificial Intelligence of Things, AIoT 2025
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 3 December 2025 through 5 December 2025
ER -