An ensemble learning based IDS using Voting rule: VEL-IDS

Area of Science:

• Cybersecurity
• Machine Learning
• Network Security

Background:

• Intrusion Detection Systems (IDSs) are crucial for analyzing internet traffic to detect and prevent cyberattacks.
• Existing IDSs often face challenges in achieving high accuracy and efficiency in attack detection.
• The CIC-CSE-IDS2018 dataset provides a relevant benchmark for evaluating IDS performance.

Purpose of the Study:

• To develop an advanced IDS with high accuracy and efficiency.
• To leverage feature selection and ensemble learning techniques for improved IDS performance.
• To reduce detection time while maintaining high accuracy in identifying network intrusions.

Main Methods:

• Dataset reduction using feature selection techniques like Spearman's correlation, Recursive Feature Elimination (RFE), and chi-square tests.
• Implementation of an ensemble learning approach combining multiple classifiers (e.g., extra trees, decision trees, logistic regression).
• Training and testing the IDS model on the CIC-CSE-IDS2018 dataset.

Main Results:

• Identification of key features that significantly enhance IDS performance.
• Demonstrated superior accuracy and reduced detection time compared to individual classifier approaches.
• Ensemble learning models achieved high accuracy rates, validating their effectiveness.

Conclusions:

• The developed IDS, utilizing feature selection and ensemble learning, offers a significant advancement in intrusion detection.
• The findings highlight the potential of ensemble methods to improve the accuracy and efficiency of IDSs.
• This research contributes to enhancing overall computer security and safeguarding against cyber threats.