E-Commerce Data Anomaly Detection Method Based on Variational Autoencoder

With the rapid development of e-commerce, data security issues on e-commerce platforms have become more and more prominent, and anomalous transactions and fraudulent behaviors have brought great challenges to the stable operation of the platform. Traditional anomaly detection methods often show certain limitations when dealing with complex and largescale data. In order to solve this problem, this paper proposes an e-commerce data anomaly detection method based on variational autoencoder (VAE). VAE is able to effectively extract features and identify anomalous data by learning the potential distribution of the data. In terms of model design, this paper introduces an adaptive threshold adjustment mechanism, combined with a lightweight classification network, which improves the detection accuracy and robustness of different types of anomalies. Experimental results show that this paper's method outperforms traditional methods in several performance indicators, especially in key indicators such as precision rate, recall rate, F1 score and ROC-AUC. The method in this paper not only helps to improve the security of e-commerce platforms, but also provides a reference for anomaly detection in other fields.