Advanced Persistent Threat Detection and Defense Algorithm Based on Multi-source Heterogeneity

Due to the Advanced Persistent Threat (APT), the traditional network threat detection methods cannot effectively find APT attacks, and it is even more difficult to achieve defense. Therefore, this paper proposes an Advanced Persistent Threat detection and defense algorithm based on multi-source heterogeneity. Firstly, the method uses machine-learning advanced persistent threat to train and output relevant alarm sequences, and then sets APT attacks as hidden state samples, and models and analyzes the correlation with the output results. Finally, the discriminant network is used for fusion training to detect APT attacks and predict the attack state in the next stage, so as to improve its defense ability. The experimental results show that this method not only has higher detection accuracy than the current machine learning algorithm, but also has a stable prediction result after observation, which puts forward a new solution for the detection and defense of modern advanced persistent threats.