Scholarlink

With the progress of blockchain technology, the adoption of smart contracts has significantly surged in several industries, making Ethereum the leading platform for smart contracts. Nevertheless, the occurrence of frequent smart contract vulnerabilities has resulted in substantial economic losses. Therefore, it is imperative to investigate efficient technologies for detecting smart contract vulnerabilities. Current methods are unable to effectively utilize the grammatical information of the contract source code in order to identify vulnerabilities in smart contracts. This work presents SCDefender, a vulnerability detection tool based on deep learning that effectively addresses the issue of smart contract re-entrancy. SCDefender employs spatial block and spatial correlation learning techniques to enhance its efficiency in detecting vulnerabilities. The system uses tree-based convolutional neural networks to detect defects in the abstract syntax tree representation of Solidity source code. Pruning technology is employed to remove unnecessary nodes and preserve essential data in the abstract syntax tree. SCDefender is a highly promising solution for detecting vulnerabilities in smart contracts due to its integration of spatial associative learning and spatial blockchain technologies. SCDefender demonstrated strong performance, achieving precision, recall, and F1 scores of 81.43%, 92.12%, and 86.45%, respectively. In addition, sensitivity research demonstrates that the vulnerability detection performance of SCDefender significantly improves as a result of implementing the abstract syntax tree pruning approach.

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

Optimizing IoT Task Offloading with Federated Deep Reinforcement Learning in Vehicle and Industrial Environments

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Tanisha Joshi,
  • G. Lakshmi Bhavani,
  • Kottala Sri Yogi,
  • Jitendra Kumar Chaudhary

摘要

With the progress of blockchain technology, the adoption of smart contracts has significantly surged in several industries, making Ethereum the leading platform for smart contracts. Nevertheless, the occurrence of frequent smart contract vulnerabilities has resulted in substantial economic losses. Therefore, it is imperative to investigate efficient technologies for detecting smart contract vulnerabilities. Current methods are unable to effectively utilize the grammatical information of the contract source code in order to identify vulnerabilities in smart contracts. This work presents SCDefender, a vulnerability detection tool based on deep learning that effectively addresses the issue of smart contract re-entrancy. SCDefender employs spatial block and spatial correlation learning techniques to enhance its efficiency in detecting vulnerabilities. The system uses tree-based convolutional neural networks to detect defects in the abstract syntax tree representation of Solidity source code. Pruning technology is employed to remove unnecessary nodes and preserve essential data in the abstract syntax tree. SCDefender is a highly promising solution for detecting vulnerabilities in smart contracts due to its integration of spatial associative learning and spatial blockchain technologies. SCDefender demonstrated strong performance, achieving precision, recall, and F1 scores of 81.43%, 92.12%, and 86.45%, respectively. In addition, sensitivity research demonstrates that the vulnerability detection performance of SCDefender significantly improves as a result of implementing the abstract syntax tree pruning approach.