Optimized extreme learning machines with deep learning for high-performance network traffic classification.

Area of Science:

• Computer Science
• Cybersecurity
• Machine Learning

Background:

• Increasing network users and applications necessitate advanced network security solutions.
• Network traffic analysis is crucial for identifying malicious activities and ensuring system integrity.
• Existing methods require enhancement for improved precision in classifying network traffic.

Purpose of the Study:

• To propose a novel framework for network traffic classification using an Improved Extreme Learning Machine (IELM).
• To enhance classification precision by optimizing model parameters and prioritizing feature relevance.
• To provide a robust and scalable solution for detecting malicious activities and mitigating security risks.

Main Methods:

• Developed an Improved Extreme Learning Machine (IELM) framework for network traffic classification.
• Integrated particle swarm optimization for optimizing IELM model parameters.
• Employed a deep learning-based feature selection mechanism to assess input feature relevance.

Main Results:

• The IELM framework demonstrated high accuracy in classifying network traffic as secure or insecure.
• Achieved a remarkable detection accuracy of 98.756% on the CICIDS 2017 dataset.
• The feature selection mechanism effectively prioritized relevant input features, enhancing classification precision.

Conclusions:

• The proposed IELM-based approach is highly effective for accurate network traffic classification.
• The framework offers a significant advancement in detecting malicious activities and strengthening network protection.
• The findings highlight the potential of IELM for robust and scalable cybersecurity solutions.