DYNAMITE: Dynamic Defense Selection for Enhancing Machine Learning-based Intrusion Detection Against Adversarial Attacks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Jing, Gungor, Onat, Shang, Zhengli, Li, Elvin, Rosing, Tajana
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915249840455680
author Chen, Jing
Gungor, Onat
Shang, Zhengli
Li, Elvin
Rosing, Tajana
author_facet Chen, Jing
Gungor, Onat
Shang, Zhengli
Li, Elvin
Rosing, Tajana
contents The rapid proliferation of the Internet of Things (IoT) has introduced substantial security vulnerabilities, highlighting the need for robust Intrusion Detection Systems (IDS). Machine learning-based intrusion detection systems (ML-IDS) have significantly improved threat detection capabilities; however, they remain highly susceptible to adversarial attacks. While numerous defense mechanisms have been proposed to enhance ML-IDS resilience, a systematic approach for selecting the most effective defense against a specific adversarial attack remains absent. To address this challenge, we propose Dynamite, a dynamic defense selection framework that enhances ML-IDS by intelligently identifying and deploying the most suitable defense using a machine learning-driven selection mechanism. Our results demonstrate that Dynamite achieves a 96.2% reduction in computational time compared to the Oracle, significantly decreasing computational overhead while preserving strong prediction performance. Dynamite also demonstrates an average F1-score improvement of 76.7% over random defense and 65.8% over the best static state-of-the-art defense.
format Preprint
id arxiv_https___arxiv_org_abs_2504_13301
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DYNAMITE: Dynamic Defense Selection for Enhancing Machine Learning-based Intrusion Detection Against Adversarial Attacks
Chen, Jing
Gungor, Onat
Shang, Zhengli
Li, Elvin
Rosing, Tajana
Cryptography and Security
The rapid proliferation of the Internet of Things (IoT) has introduced substantial security vulnerabilities, highlighting the need for robust Intrusion Detection Systems (IDS). Machine learning-based intrusion detection systems (ML-IDS) have significantly improved threat detection capabilities; however, they remain highly susceptible to adversarial attacks. While numerous defense mechanisms have been proposed to enhance ML-IDS resilience, a systematic approach for selecting the most effective defense against a specific adversarial attack remains absent. To address this challenge, we propose Dynamite, a dynamic defense selection framework that enhances ML-IDS by intelligently identifying and deploying the most suitable defense using a machine learning-driven selection mechanism. Our results demonstrate that Dynamite achieves a 96.2% reduction in computational time compared to the Oracle, significantly decreasing computational overhead while preserving strong prediction performance. Dynamite also demonstrates an average F1-score improvement of 76.7% over random defense and 65.8% over the best static state-of-the-art defense.
title DYNAMITE: Dynamic Defense Selection for Enhancing Machine Learning-based Intrusion Detection Against Adversarial Attacks
topic Cryptography and Security
url https://arxiv.org/abs/2504.13301