CYBER THREAT DETECTION BASED ON ARTIFICIAL NEURAL NETWORKS

A significant challenge in cybersecurity is automating cyber threat detection efficiently. Our proposed method converts multiple security events into individual profiles and utilizes deep learning for enhanced identification within an AI-SIEM system. Integrating event profiling with various neural network methods like FCNN, CNN, and LSTM distinguishes true positives from false positives, enabling rapid threat response. Benchmark experiments (NSLKDD, CICIDS 2017) and real-world datasets compared our approach with five traditional machine learning techniques. Results indicate our method’s effectiveness for network intrusion detection, even outperforming conventional methods in real-world scenarios.