Examination and Performance Evaluation of Smart Contract Reentrancy Vulnerability using Deep Learning Detection Tools

Introduction: This study proposes, from the perspective of optimizing the model feature space, a deep learning-based vulnerability detection technique for Ethereum smart contracts based on Raptor vision invasive Hunt optimization. Objectives: To examine experimental results of available toools and novel model in detecting the re-entrancy vulnerability. Methods: By combining the coordinated invasive hunting behaviour of Gallus domesticus with the intelligent vision-based in-depth driving behaviour of osprey, the RVIhO algorithm is achieved. They successfully adjust the model's hyperparameters with increased productivity for precise vulnerability detection based on these hybridising traits. Results: A large-scale dataset with different vulnerabilities having the findings demonstrate that it suggested strategy performs exceptionally well in terms of detection 93.25% accuracy rates attained for detection of re-entrancy vulnerabilities of provided smart contract data set. Conclusions: The results presented, and performance compared with the best available deep learning methods for vulnerability detection.