Year: 2025
Type: conference-paper
Abstract: In order to improve the financial information management ability of listed companies, this paper constructed an anomaly detection model combining Principal Component Analysis (PCA) and Variational Auto-Encoders (VAE). Based on the collection and preprocessing of financial data, PCA is used for feature extraction, VAE is used to learn the potential distribution of data, and k-fold cross validation is used to improve the prediction performance of the model. Experimental results show that when the ratio of training set to test set is 9:1, PCA-VAE shows excellent performance in anomaly detection tasks, with the accuracy, recall and F1 score of 0.9712, 0.9656 and 0.9688, respectively, which are significantly better than the traditional machine learning algorithm and the classification model combined with PCA method.
Author: Yuchen Jiang
Institution: Hohai University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4413601762