SCGformer: Smart contract vulnerability detection based on control flow graph and transformer

Abstract The security of smart contract has always been one the significant problems in blockchain. As shown previous studies, vulnerabilities contracts can lead to unpredictable losses. With rapid growth number contracts, more and data driven detection technologies based on machine learning have proposed. However, some state‐of‐the‐art approaches mainly rely source code contract. These methods are limited by openness version programming language. To address this problem, we propose a novel vulnerability method transformer constructing control flow graph (CFG) operation codes (opcodes), which shields difference various versions program Extensive experiments conducted evaluate effectiveness proposed authors' own collected dataset. experimental results show that achieves 94.36% accuracy detection, performs better than other methods.