CSGraph2Vec: A Distributed Representation Learning Approach for Assembly Functions
Wael J. Alhashemi · eScholarship@McGill (McGill) · 2023
Software reverse engineering is an essential yet time-consuming undertaking in the identification of malware, software vulnerabilities, and plagiarism, especially when access to the source code is limited. Due to the development of machine as well as deep learning, automating the construction of vector embeddings has grown more feasible. This research introduces CSGraph2Vec, a distributed and automated deep learning approach that produces representations of assembly functions. By leveraging the power of the Electra pre-trained language model, as well as message-passing neural networks, CSGraph2Vec efficiently incorporates control flow and semantic information from assembly code. Our model successfully learns significant features that distinguish benign from malicious functions. Through extensive experimentation and evaluation of the malware classification task, we show that our model performs better than several alternative approaches