AN EVENT PROFILE–BASED DEEP NEURAL NETWORK FRAMEWORK FOR CYBERSECURITY APPLICATIONS

Abstract

The rapid evolution of cyber threats poses significant challenges to traditional security systems, which often fail to adapt to the dynamic nature of attack patterns. To address this gap, this study introduces NeuroShield, a deep neural network (DNN)–based framework that leverages event profiling for advanced cyber threat detection. By extracting and analyzing event profiles from system logs, network activities, and user behavior, the model learns complex patterns associated with malicious activities. The proposed approach integrates feature engineering with deep learning techniques to improve detection accuracy, reduce false alarms, and enhance real-time responsiveness. Experimental results demonstrate that NeuroShield outperforms conventional machine learning classifiers, achieving higher precision and recall in identifying both known and novel cyber threats. The findings highlight the potential of event profile–driven DNNs as a robust solution for modern cybersecurity, enabling proactive defense mechanisms and adaptive resilience in critical digital infrastructures.