Abstract:Unsupervised anomaly detection (AD) methods, either reconstruction based or prediction based, determine anomalies based on residuals. Occasional mutations in a single variable can cause the residuals to exceed the limits. Indeed, such mutations are not variations in the operating mechanism of the system. Thus, system-level anomalies are challenging to characterize. Complex networks (or "graphs") are well adapted for modeling and characterizing the laws of evolution of complex systems. However, most industrial scenarios are without graphs. Hence, a self-supervised variational graph autoencoders (SS-VGAE) method is proposed. First, the multisource sensor dynamic graph is constructed through detrended cross-correlation analysis (DCCA). Second, target and self-supervised learning tasks are designed. The target task is to reconstruct the input graph structure to minimize the reconstruction loss. The self-supervised task is to learn the optimal center of the hypersphere in the latent space such that the mean features are gathered toward the center as much as possible. Multitask joint optimization allows high- and low-dimensional space features to be considered simultaneously, thereby improving the reliability of anomaly scores. Then, the distribution of anomaly scores is calculated and integrated into a system health indicator (HI). The system HI is more applicable to assist decision making. Finally, the superiority of the proposed method, namely, better detection accuracy and robustness, is demonstrated by nuclear personal computer transient analyzer (PCTRAN) simulation data and Skoltech anomaly benchmark (SKAB) data. Last but not least, systematic anomalies are found to make the correlation between the variables stronger.

Fast Particle-based Anomaly Detection Algorithm with Variational Autoencoder

Robust and Unsupervised KPI Anomaly Detection Based on Highly Sensitive Conditional Variational Auto-Encoders.

Resilient VAE: Unsupervised Anomaly Detection at the SLAC Linac Coherent Light Source

VESC: a new variational autoencoder based model for anomaly detection

Variational Autoencoder for Anomaly Detection: A Comparative Study

Anomaly Detection of Time Series With Smoothness-Inducing Sequential Variational Auto-Encoder

Unsupervised Anomaly Detection Using Variational Auto-Encoder based Feature Extraction

TeVAE: A Variational Autoencoder Approach for Discrete Online Anomaly Detection in Variable-state Multivariate Time-series Data

Multi-mode non-Gaussian variational autoencoder network with missing sources for anomaly detection of complex electromechanical equipment

Time Series Anomaly Detection with Variational Autoencoders

Unsupervised anomaly video detection via a double-flow ConvLSTM variational autoencoder

VELC: A New Variational AutoEncoder Based Model for Time Series Anomaly Detection

GAN-AE : An anomaly detection algorithm for New Physics search in LHC data

Autoencoders for unsupervised anomaly detection in high energy physics

Improved Anomaly Detection through Conditional Latent Space VAE Ensembles

Learning Sparse Representation With Variational Auto-Encoder for Anomaly Detection.

Self-Supervised Variational Graph Autoencoder for System-Level Anomaly Detection

Anomaly Detection for Time Series Using VAE-LSTM Hybrid Model

Anomaly Detection in Particle Accelerators using Autoencoders

StackVAE-G: an Efficient and Interpretable Model for Time Series Anomaly Detection

Particle-based fast jet simulation at the LHC with variational autoencoders