제안 방법

• In this paper, we proposed a permission based approach to find out the malicious method and automatically repair the malicious code using LSTM technique. As the growth rate of Android malware, millions apps produced in Google Play.
• To this end, the system builds two dataset according to the unique permission values and Intents retrieved from training dataset. Then, the LSTM adopted to embed the semantics of Intents and build the relevance among permission vector and Intent vectors. As a result, the malicious method can be classified by Softmax.