AI-Driven Threat Detection and Response in Cybersecurity Using Autonomous Adaptive Approach

The exacerbating complexity and frequency of cyber threats present notable obstacles to conventional cybersecurity measures, necessitating the creation of more dynamic and intelligent systems. In this paper we developed a hybrid autonomous adaptive AI threat detection framework using hybrid machine learning algorithms such as unsupervised learning, supervised learning and reinforcement learning. The unsupervised learning is use for anomaly detection, supervised learning for threat classification and reinforcement learning for autonomous decision making. The system was implemented and analyse using NSL-KDD cybersecurity datasets to continuously learn from evolving attack pattern and autonomously respond to mitigate cyber threat risks in real time. The analysis result shows that the hybrid framework achieved 96.8% accuracy, 95.4 % precision, 97.2% recall, 93.6% F1-Score, 2.1% FPR and response time of 25ms. The result indicates that the hybrid framework achieved a strong learning ability in correctly identifying attacks, minimized the number of false threat alert, reduced system workload during analysis and speedily mitigate real-time threats detected in live network.