TransFront: Bi-path Feature Fusion for Detecting Front-running Attack in Decentralized Finance

The rise of decentralized finance (DeFi) has brought about an unprecedented wave of financial innovation on blockchain platforms like Ethereum. However, this growth has also exposed vulnerabilities, particularly front-running attacks, where malicious actors exploit transaction ordering to gain illegal profits. We present TransFront, a novel approach for detecting transaction-based front-running attacks in DeFi. By leveraging a bi-path feature fusion strategy, TransFront adeptly captures both macro and micro transaction patterns, improving the accuracy and reliability of detection. This method addresses the limitations of existing detection techniques and provides a comprehensive solution by integrating feature extraction, feature fusion, and classification into an end-to-end model. Experimental results demonstrate that TransFront outperforms several state-of-the-art methods, achieving superior AUC, precision, recall, and F1-score performance. These findings underscore the potential of TransFront to significantly improve transaction security and asset protection in DeFi.