Autonomous Cyber Defense: An AI-Based SelfHealing Framework
NeuroShield is an intelligent, machine learning– based Intrusion Detection and Response System designed to detect and mitigate modern real-time cyberattacks. The system uses live network packet capture, flow-based feature extraction aligned with the UNSW-NB15 dataset, and a Random Forest classifier trained on the top 10 most informative features across 10 attack categories. A multi-stage pipeline enables continuous traffic monitoring, accurate attack classification, and automated mitigation through a custom healing engine. Upon detection, NeuroShield executes targeted responses such as IP blocking, session termination, and alert logging. Validation using simulated attack scenarios confirmed high detection accuracy, low latency, and reliable real-time defense performance.