An Artificial Intelligence Method to Predict Malicious Behavior

Abstract

With the advancement of technology, more and more information equipment appear in people’s lives. Up to date, with the improvement of network technology, the transmission of information between devices has become more convenient and faster, and the worries of information security follow. Although discussion of the information security of terminal equipment can be an issue, the information of terminal equipment will eventually be sent back to the server. Therefore, the research of the intrusion detection of servers is more fundamental. It is well known that the appearance of malicious behavior often means that the system may have been attacked by hackers. Thus, early detection of malicious behavior plays a vital role in preventing hackers from intrusion. However, most of the current known researches tend to focus only on how the system recognizes the malicious behavior when it is occurred, but the system cannot predict the occurrence in advance when the malicious behavior has not been completed. This research hopes to propose a method that can predict the appearance of malicious behavior before the malicious behavior is completed. We propose a method for predicting malicious behavior, which can determine whether the behavior is malicious before it is completed. The method of this research is to construct a malicious behavior prediction model by GAN (Generative Adversarial Network). It is based on the malicious behavior detection model established by the LSTM (Long Short-Term Memory) model. The experimental results show that the prediction accuracy of the model is about 83%.