SA-DRL: Security-Aware Deep Reinforcement Learning for Ransomware Detection with Asymmetric Reward Design

Ransomware detection is a security-critical task in which false negatives and false positives have unequal operational consequences. Conventional machine learning detectors often use symmetric objectives that penalize missed ransomware detections and benign false alarms equally, although a false neg...