Digital Currency Abnormal Detection Based on Graph Neural Network

In recent years, with the rapid development of the digital currency market, the security issues of digital currency transactions have become increasingly prominent, and abnormal transaction detection has become an important technical means to ensure the security of the financial system. Traditional detection methods face many challenges in dealing with complex blockchain data, as they cannot fully utilize the inherent structural information of transaction data, and are difficult to handle the problem of imbalanced categories. Accordingly, this study proposes a GNN (Graph Neural Network)-based abnormal transaction detection method, which models transaction data as a graph structure, extracts transaction features using Graph Convolutional Network (GCN) and uses small sample processing techniques to alleviate data imbalance. The proposed method showed that the experimental results could effectively improve the detection performance of abnormal transactions after multiple rounds of training, especially in terms of AUC evaluation indicators, which verifies its potential application in the detection of abnormal transactions in digital currencies. This study provides a novel and effective solution for anomaly detection in complex network data and points out the direction for future research.