An intelligent cyber security phishing detection system using deep learning techniques.

Area of Science:

• Cybersecurity
• Machine Learning Applications
• Social Engineering

Background:

• Phishing attacks represent a significant and escalating threat to individuals, governments, and businesses.
• Existing detection methods are insufficient against sophisticated phishing techniques, particularly email-based scams.
• There is a critical need for advanced technologies to mitigate the growing problem of phishing emails.

Purpose of the Study:

• To provide a comprehensive overview of machine learning (ML) in the context of cybersecurity.
• To analyze common phishing attack techniques and identify the most effective methods used by attackers.
• To propose and evaluate an ML-based model for detecting phishing emails.

Main Methods:

• A survey of current phishing attack vectors, with a focus on email-based threats.
• Development of a phishing email detection model using ML techniques.
• Training and validation of the ML model using three distinct datasets, incorporating email text and other relevant features.

Main Results:

• The study found that utilizing a larger number of features in the ML model leads to more accurate and efficient phishing detection.
• The boosted decision tree algorithm demonstrated high performance across the applied datasets.
• The best ML algorithm achieved accuracies of 0.88, 1.00, and 0.97 on the respective datasets.

Conclusions:

• Machine learning offers a powerful approach to enhance the detection of phishing emails.
• Feature engineering is crucial for maximizing the effectiveness of ML-based phishing detection systems.
• The proposed model and findings contribute to developing more robust defenses against phishing threats.