Semantic Awareness Model For Binary Operator Code

With the promotion of artificial intelligence in various industries, there have been some organizations and individuals using various means to attack it. Common types of attacks against models include adversarial sample attacks, data poisoning attacks, and model stealing attacks. The above attack methods require a certain understanding of the structure of the model, so it becomes a challenge to restore the category of binary operators from the model.Binary code similarity detection (BCSD) has important applications in code checking, vulnerability detection, and malicious co de analysis. Due to the lack of syntax structure information in binary operator code, it is difficult to determine the type of operator. Recent research has focused on using deep learning models to understand the semantics and structural information of binary code to achieve better results. Recent research has shown that deep learning models, especially natural language processing models, can comprehend the semantics of binary code. In this paper, we propose a method for identifying binary operators which uses a sequence-aware model and a structure-aware model to model binary operators. It combines CNN semantics and Transformer semantics for classification. The evaluation shows that our method can achieve good performance in binary operator classification.