A Hybrid Framework for Ransomware Detection Using Deep Learning and Monte Carlo Tree Search
Guan Li, Shaohui Wang, Yanbin Chen, Jie Zhou, Qihang Zhao · 2024
Ransomware attacks have emerged as a dominant cybersecurity threat, with increasingly sophisticated techniques that often evade traditional detection methods. A novel framework is proposed that synergizes the predictive strengths of deep learning models with the dynamic decision-making capabilities of Monte Carlo Tree Search (MCTS), providing a comprehensive solution to the challenges posed by evolving ransomware variants. Through rigorous evaluation, the hybrid framework demonstrated a significant improvement in detection accuracy while reducing false positives, outperforming conventional machine learning models. The integration of MCTS allowed for the exploration of multiple decision paths, enhancing the system’s adaptability to novel threats in real time. Additionally, the proposed model maintained computational efficiency, making it feasible for real-time deployment in enterprise environments. The results demonstrate the hybrid model's potential as a robust defense mechanism in modern cybersecurity, offering a scalable and efficient tool for mitigating ransomware threats.