A novel hybrid feature selection and ensemble-based machine learning approach for botnet detection.

Area of Science:

• Cybersecurity
• Machine Learning
• Network Security

Background:

• Botnet detection is a critical challenge due to evolving cyber threats.
• Existing detection systems struggle against advanced botnet strategies.
• A dynamic and proactive approach is necessary for effective botnet detection.

Purpose of the Study:

• To develop a novel solution for botnet detection by combining advanced feature selection and ensemble machine learning.
• To enhance the accuracy and adaptability of botnet detection systems.
• To establish a new benchmark for reliability in identifying botnet activities.

Main Methods:

• Hybrid Feature Selection: Categorical Analysis, Mutual Information, and Principal Component Analysis were combined.
• Ensemble Machine Learning: Extra Trees classifier was employed as the primary technique.
• Input Space Refinement: Feature selection methods were used to optimize the data for the ensemble learner.

Main Results:

• Achieved a near-perfect accuracy rate of 99.99% in botnet classification.
• Demonstrated exceptional adaptability to new and evolving botnet phenomena.
• Consistently achieved over 99% True Positive Rates with a False Positive Rate near 0.00%.

Conclusions:

• The proposed hybrid model offers unprecedented precision and resilience in botnet detection.
• This research provides a transformative step in cybersecurity, setting a new precedent for reliability.
• The findings offer a blueprint for developing next-generation security frameworks against botnet infiltrations.