Computational Intelligence Approaches in Developing Cyberattack Detection System.

Area of Science:

• Cybersecurity
• Network Security
• Machine Learning Applications

Background:

• Intrusions compromise system integrity, privacy, and accessibility.
• The Internet of Things (IoT) and networks are vulnerable to various cyberattacks.
• Effective intrusion detection systems are crucial for maintaining digital security.

Purpose of the Study:

• To develop and validate an intelligent cybersecurity system for detecting network intrusions.
• To evaluate the performance of machine learning and deep learning algorithms for intrusion detection.
• To enhance the security of Internet of Things (IoT) devices and networks.

Main Methods:

• Applied machine learning algorithms: Quantum Support Vector Machine (QSVM), k-nearest neighbor (KNN), Long Short-Term Memory (LSTM), and autoencoders.
• Utilized correlation method for network feature selection, identifying nine key features.
• Employed one-hot encoding for categorical feature conversion and validated using the KDD Cup database.

Main Results:

• KNN and LSTM algorithms demonstrated superior performance in classifying network traffic as normal or attack.
• Achieved high accuracy for binary classification (KNN: 98.55%, LSTM: 97.28%) and multiple classifications (KNN: 98.28%, LSTM: 97.07%).
• Identified KNN and LSTM as suitable algorithms for building effective intrusion detection systems.

Conclusions:

• The proposed intelligent cybersecurity system effectively detects network intrusions.
• KNN and LSTM algorithms are highly effective for intrusion detection, offering robust classification performance.
• The study highlights the potential of machine learning in securing IoT devices and networks against cyber threats.