STAD-GAN: Unsupervised Anomaly Detection on Multivariate Time Series with Self-training Generative Adversarial Networks

Abstract

Anomaly detection on multivariate time series (MTS) is an important research topic in data mining, which has a wide range of applications in information technology, financial management, manufacturing system, etc. However, the state-of-the-art unsupervised deep learning models for MTS anomaly detection are vulnerable to noise and have poor performance on the training data containing anomalies. In this paper, we propose a novel Self-Training based Anomaly Detection with Generative Adversarial Network (GAN) model called STAD-GAN to address the practical challenge. The STAD-GAN model consists of a generator-discriminator structure for adversarial learning and a neural network classifier for anomaly classification. The generator is learned to capture the normal data distribution, and the discriminator is learned to amplify the reconstruction error of abnormal data for better recognition. The proposed model is optimized with a self-training teacher-student framework, where a teacher model generates reliable high-quality pseudo-labels to train a student model iteratively with a refined dataset, so that the performance of the anomaly classifier can be gradually improved. Extensive experiments based on six open MTS datasets show that STAD-GAN is robust to noise and achieves significant performance improvement compared to the state-of-the-art.

Anomaly detection in multivariate time series data using deep ensemble models

“…The DVGCRN paradigm incorporates spatial and temporal dependency in MTS data by merging deep variational inference with graph convolutional recurrent networks. Despite [ 35 ] concerns with computer complexity and hyperparameter adjustment, the investigation shows that this strategy dramatically improves anomaly identification performance in MTS data. The STAD-GAN architecture converts the given MTS into a latent model, subsequently fed into a deep neural network classifier.…”

Management Analysis Method of Multivariate Time Series Anomaly Detection in Financial Risk Assessment

“…This adversarial training process continually enhances the realism of the generated data by improving the generator's ability, while the discriminator becomes increasingly adept at differentiation. In the context of time series anomaly detection, the generative capability of GANs positions them as a promising tool for learning and generating the distribution of normal time series data to detect anomalous patterns (Zhang et al, 2023).…”