Smart contract vulnerability detection combined with multi-objective detection

Abstract

Blockchains have been booming in recent years. As a decentralized system architecture, smart contracts give blockchains a user-defined logic. A smart contract is an executable program that can automatically carry out transactions on the Ethereum blockchain. However, some security issues in smart contracts are difficult to fix, and smart contracts also lack quality assessment standards. Therefore, this study proposes a Multiple-Objective Detection Neural Network (MODNN), a more scalable smart contract vulnerability detection tool. MODNN can validate 12 types of vulnerabilities, including 10 recognized threats, and identify more unknown types without the need for specialist or predefined knowledge through implicit features and Multi-Objective detection (MOD) algorithms. It supports the parallel detection of multiple vulnerabilities and has high scalability, eliminating the need to train separate models for each type of vulnerability and reducing significant time and labor costs. This paper also developed a data processing tool called Smart Contract-Crawler (SCC) to address the lack of smart contract vulnerability datasets. MODNN was evaluated using more than 18,000 smart contracts from Ethereum. Experiments showed that MODNN could achieve an average F1 Score of 94.8%, the current highest compared to several standard machine learning (ML) classification models.