.png){kind=icon}
.png){kind=icon}
Return
Enhancing Performance of Intrusion Prevention Systems through Machine Learning: A Comparative Study
.png){kind=icon}
PRE
AI
A
K
DOI：10.21123/2411-7986.5222.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This research introduces a Machine Learning-Powered Intrusion Prevention System (ML-IPS) as a robust solution to address the challenges posed by evolving cyber threats. The ML-IPS combines timely threat detection with enhanced accuracy for real-time attack prevention, offering a resilient defense against a broad spectrum of cyber-attack. This study delves into the comprehensive evaluation of a machine learning-powered IPS that ingeniously harnesses the power of advanced algorithms to facilitate real-time threat detection and significantly enhance the overall accuracy of the system. The efficacy of intrusion prevention systems (IPS) that employ machine learning (ML) is greatly dependent on the choice of suitable ML algorithms and the evaluation of their precision and inference duration. This research embarks on an in-depth evaluation of ML models for IPS applications, focusing on a comprehensive comparison of their accuracy and inference time metrics. Additionally, the methodology employs a supervised learning approach using a labeled dataset (CICIDS2017) containing both benign and malicious network traffic, providing a realistic and practical approach. The simulation results demonstrate that Decision tree and random forest algorithms can improve the prevention of attack in real-time by achieving 99.88% accuracy and about 10ms for time of detection. The findings demonstrate that the Intrusion Prevention System (IPS) is adept at promptly identifying and reacting to assaults, thereby furnishing a stronger and more durable safeguard against the ever-changing landscape of cyber hazards.
Keywords:
Cybersecurity
Decision tree
KNN
Machine learning
Random forest
Journal
B
IF:
1.7
Papers:
120
Citations:
1.9K
Organization
Cited Papers