Scholarlink

Ethereum being a dominant player in decentralized finance, invites a vast number of smart contracts annually. Though smart contracts play a crucial role in developing decentralized applications that are safe, secure, and efficient, Ethereum lost a huge amount due to various attacks caused by different vulnerabilities within the smart contracts. Therefore, it is crucial to audit the reliability of smart contracts before their deployment. This paper proposes SmartAudit which identifies whether the smart contract is vulnerable. SmartAudit employs BERT (Bidirectional Encoder Representations from Transformers) to identify vulnerabilities in unseen smart contracts. It utilizes optimized opcode sequences derived from a pruned control flow graph as input. The results show that SmartAudit significantly improves accuracy, precision, recall and F1-score compared to the state-of-the-art models.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

SmartAudit: Smart Contract Vulnerability Detection Using Transfer Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ankur Jain,
  • Somanath Tripathy

摘要

Ethereum being a dominant player in decentralized finance, invites a vast number of smart contracts annually. Though smart contracts play a crucial role in developing decentralized applications that are safe, secure, and efficient, Ethereum lost a huge amount due to various attacks caused by different vulnerabilities within the smart contracts. Therefore, it is crucial to audit the reliability of smart contracts before their deployment. This paper proposes SmartAudit which identifies whether the smart contract is vulnerable. SmartAudit employs BERT (Bidirectional Encoder Representations from Transformers) to identify vulnerabilities in unseen smart contracts. It utilizes optimized opcode sequences derived from a pruned control flow graph as input. The results show that SmartAudit significantly improves accuracy, precision, recall and F1-score compared to the state-of-the-art models.