A Deep Reinforcement Learning Malware Detection Method Based on PE Feature Distribution
Binxiang Liu, Gang Zhao, Ruoying Sun · 2019
Existing anti-virus software and malware detection methods, including signature-based and the machine learning-based malware detection methods, are unable to update the virus database in real time, resulting in poor resistance to malware variants. To solve this problem, this paper proposes a novel malware detection method based on deep reinforcement learning, which combines the advantages of Q-learning and neural network. Q-learning action selection strategy is adopted while solving the problem of high dimensional state space. Theoretical analysis and experimental results show that the proposed method can not only detect malware variants efficiently, but also perform well in many well-known anti-virus software, which is a new direction in the field of malware detection.