This paper delves into the area of binary function similarity detection, which is crucial in various security applications. The researchers construct a cross-inlining dataset and discover three cross-inlining patterns. They then propose a new model, CI-Detector, which uses the attributed CFG to represent the semantics of binary functions and GNN to embed binary functions into vectors. The model is then trained for these three cross-inlining patterns. The results show that CI-Detector can detect cross-inlining pairs with a precision of 81% and a recall of 97%, surpassing all existing methods.
Publication date: 15 Jan 2024
Project Page: https://doi.org/10.1145/3597503.3639080
Paper: https://arxiv.org/pdf/2401.05739
