Exploring Deep Learning for Vulnerability Detection in Smart Contracts

Abstract

This project explores vulnerability detection in Solidity smart contracts. The following report provides a brief overview of blockchain technology, smart contract specific vulnerabilities and the tooling that exists to detect these vulnerabilities. The application of deep learning as a vulnerability detection tool was explored in more detail. The result of this work is an LSTM trained to detect re-entrancy vulnerabilities in smart contracts. The model is trained on smart contracts identified and labeled in the ScawlD dataset provided by Yashavant et al.