Quantum particle Swarm optimized extreme learning machine for intrusion detection

Abstract

Ensuring a secure online environment hinges on the timely detection of network attacks. Nevertheless, existing detection methods often grapple with the delicate balance between speed and accuracy. In this paper, we introduce a novel intrusion detection algorithm that marries quantum particle swarm optimization with an extreme learning machine (QPSO-ELM). Firstly, we present a feature selection algorithm grounded in partitioned gains to distill vital features from data samples, thereby diminishing feature count to amplify both model training speed and accuracy. Subsequently, we unveil an intrusion detection scheme underpinned by QPSO-ELM, capable of achieving exceptional levels of training and detection speed, all while maintaining high accuracy. Finally, we fine-tune the trained model using the proposed hidden layer node selection algorithm, reducing the detection model size without compromising detection accuracy, thus further elevating its speed. The experiment results indicate that compared to the current baseline, our proposed intrusion detection scheme achieves the best results in terms of accuracy, precision, recall, and detection latency. Furthermore, the ablation experiment results demonstrate the effectiveness of our proposed method in improving both detection speed and detection accuracy.

Full text

Vol:.(1234567890)
The Journal of Supercomputing (2024) 80:14622–14644
https://doi.org/10.1007/s11227-024-06022-y
1 3
Quantum particle Swarm optimized extreme learning
machine forintrusion detection
HanQi1· XinyuLiu1· AbdullahGani2· ChangqingGong1
Accepted: 23 February 2024 / Published online: 21 March 2024
© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature
2024
Abstract
Ensuring a secure online environment hinges on the timely detection of network
attacks. Nevertheless, existing detection methods often grapple with the delicate
balance between speed and accuracy. In this paper, we introduce a novel intru-
sion detection algorithm that marries quantum particle swarm optimization with
an extreme learning machine (QPSO-ELM). Firstly, we present a feature selection
algorithm grounded in partitioned gains to distill vital features from data samples,
thereby diminishing feature count to amplify both model training speed and accu-
racy. Subsequently, we unveil an intrusion detection scheme underpinned by QPSO-
ELM, capable of achieving exceptional levels of training and detection speed, all
while maintaining high accuracy. Finally, we fine-tune the trained model using the
proposed hidden layer node selection algorithm, reducing the detection model size
without compromising detection accuracy, thus further elevating its speed. The
experiment results indicate that compared to the current baseline, our proposed
intrusion detection scheme achieves the best results in terms of accuracy, precision,
recall, and detection latency. Furthermore, the ablation experiment results demon-
strate the effectiveness of our proposed method in improving both detection speed
and detection accuracy.
Keywords Network security· Feature selection· Quantum particle swarm
optimization· Extreme learning machine· Intrusion detection
1 Introduction
With the rapid advancement of information technology, the internet is expanding
at an unprecedented rate, becoming increasingly intertwined with our daily lives.
Unfortunately, malicious actors, often motivated by personal gain, are continually
orchestrating a wide array of network attacks, including malware attacks, exploi-
tation of zero-day vulnerabilities, web-based assaults, and disruptive Distributed
Extended author information available on the last page of the article
Content courtesy of Springer Nature, terms of use apply. Rights reserved.