Unmasking Cybercrime with Artificial-Intelligence-Driven Cybersecurity Analytics.

Area of Science:

• Computer Science
• Cybersecurity
• Artificial Intelligence

Background:

• The global rise in cybercrime, accelerated by the COVID-19 pandemic, poses a significant threat to national GDP.
• Cybercrime encompasses diverse offenses like hacking, phishing, fraud, malware, and botnet attacks.
• Existing cybersecurity measures require enhancement to realistically combat increasingly sophisticated cyber threats.

Purpose of the Study:

• To develop and evaluate a novel collaborative deep learning approach for early identification and detection of botnet attacks.
• To improve the understanding of cyber threat intelligence and emerging botnet attack vectors.
• To enhance cybersecurity forensic investigation procedures.

Main Methods:

• A hybrid deep learning model integrating unsupervised Long Short-Term Memory (LSTM) and supervised Convolutional Neural Network (CNN) was proposed.
• The model was trained and validated using established datasets: CTU-13 and IoT-23.
• Performance was assessed based on detection rates and false positive rates.

Main Results:

• The proposed deep learning approach demonstrated superior performance in botnet attack detection.
• Achieved a high success rate exceeding 98.7%.
• Maintained a low false positive rate of 0.04%.

Conclusions:

• The collaborative deep learning model effectively enhances early detection of botnet attacks.
• The study contributes to advancing cyber threat intelligence and forensic capabilities.
• This approach offers a robust solution for improving cybersecurity against evolving threats.