Home / Current Issue / Paper 1711751
Particle Swarm Optimization-Based Artificial Neural Network for Network Intrusion Detection
Subject area: Science,Engineering and Technology · Area of research: Computer Science
Abstract
The growing complexity of cyberattacks has made efficient intrusion and malware detection an urgent challenge. This study addresses this problem by integrating Particle Swarm Optimization (PSO) with Artificial Neural Networks (ANN) to enhance classification accuracy while reducing computational costs. Using the CIC-IDS2017 and EMBER 2018 benchmark datasets, PSO was employed to select the most relevant features that improve the ANN?s predictive capability. The approach was compared with models trained on all available features to evaluate the trade-off between dimensionality reduction and performance. Results showed that the PSO-selected feature model achieved higher accuracy and efficiency, with the CIC-IDS2017 dataset recording precision and accuracy values of 99.4% and 99.78%, the EMBER 2018 dataset reaching 96% accuracy. These findings demonstrate that PSO effectively eliminates redundant features, leading to faster convergence and improved generalization. The proposed PSO-ANN framework offers a scalable and robust solution for intrusion and malware detection, contributing to the advancement of intelligent cybersecurity systems.
Keywords
Particle Swarm Optimization (PSO), Artificial Neural Network (ANN), Feature Selection, Intrusion Detection
How to cite this paper
@article{1711751,
author = {Michael Manasseh Dalaky, Gideon yunius Giroh},
title = {Particle Swarm Optimization-Based Artificial Neural Network for Network Intrusion Detection},
journal = {Iconic Research And Engineering Journals},
year = {2025},
volume = {9},
number = {5},
pages = {242-247},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1711751.pdf},
abstract = {The growing complexity of cyberattacks has made efficient intrusion and malware detection an urgent challenge. This study addresses this problem by integrating Particle Swarm Optimization (PSO) with Artificial Neural Networks (ANN) to enhance classification accuracy while reducing computational costs. Using the CIC-IDS2017 and EMBER 2018 benchmark datasets, PSO was employed to select the most relevant features that improve the ANN?s predictive capability. The approach was compared with models trained on all available features to evaluate the trade-off between dimensionality reduction and performance. Results showed that the PSO-selected feature model achieved higher accuracy and efficiency, with the CIC-IDS2017 dataset recording precision and accuracy values of 99.4% and 99.78%, the EMBER 2018 dataset reaching 96% accuracy. These findings demonstrate that PSO effectively eliminates redundant features, leading to faster convergence and improved generalization. The proposed PSO-ANN framework offers a scalable and robust solution for intrusion and malware detection, contributing to the advancement of intelligent cybersecurity systems.},
keywords = {Particle Swarm Optimization (PSO), Artificial Neural Network (ANN), Feature Selection, Intrusion Detection},
month = {November},
}