A Hybrid Neural Network Model-based Approach for Detecting Smart Contract Vulnerabilities

Smart contract vulnerabilities have become a common source of security incidents in the blockchain network in recent years. To mitigate the impact of such vulnerabilities, scholars have been exploring more effective and dependable methods for detecting them. However, existing smart contract vulnerability detection methods suffer from issues like a high rate of false positives and omissions, limited scalability, and reliance on expert knowledge, among others. To address these challenges, this study proposes a hybrid neural network model-based approach to smart contract vulnerability detection that leverages the strengths of different neural networks. By incorporating global context alongside local feature extraction, the method significantly enhances feature extraction rates. Experimental results demonstrate that the proposed method is highly efficient and accurate, making it a suitable solution for detecting smart contract vulnerabilities.