Abstract:Multivariate time series usually have entangled temporal patterns and various anomaly types. Meanwhile, they often contain both continuous and discrete features. Many existing methods directly model correlations in complex multivariate time series to conduct anomaly detection. Decomposing time series into different components, such as the overall trend and fluctuations, can contribute to better extracting semantic information and detecting anomalies. Existing decomposition-based anomaly detection methods still have several limitations. First, they directly decompose all features without considering that discrete features are unsuitable for decomposition because they do not have trends or fluctuations. Second, they adopt the same networks for different components with different characteristics, limiting their ability to extract semantic information. Moreover, due to the nature of Transformers, existing reconstruction-based methods using Transformers rarely form information bottlenecks, reducing the differentiation between the reconstruction errors of normal data and anomalies. This paper proposes a multivariate time series anomaly detection method with separation, decomposition, and dual Transformer-based autoencoder (SDDformer). Different from existing methods, SDDformer separates continuous and discrete features and only decomposes continuous features into trend and residual components. Considering the different characteristics of different components, SDDformer adopts Crossformer and the vanilla Transformer as the backbone of two different autoencoders to reconstruct the trend and residual components. Information bottlenecks are better formed using an extra token as the latent variable between the encoder and the decoder. SDDformer regards reconstructing a discrete feature as a classification task and calculates Cross-Entropy as its reconstruction error. Three different metrics are adopted in this paper to compare SDDformer with a variety of typical anomaly detection methods on public data sets, and the experimental results prove that SDDformer can achieve state-of-the-art performance.

Local-Adaptive Transformer for Multivariate Time Series Anomaly Detection and Diagnosis.

A Time Series Anomaly Detection Method Based on Series-Parallel Transformers with Spatial and Temporal Association Discrepancies

A two-stage adversarial Transformer based approach for multivariate industrial time series anomaly detection

Hybrid Graph Transformer Networks for Multivariate Time Series Anomaly Detection

Transformer-based multivariate time series anomaly detection using inter-variable attention mechanism

Multivariate Time Series Anomaly Detection with Fourier Time Series Transformer

Anomaly Transformer: Time Series Anomaly Detection with Association Discrepancy

MEMTO: Memory-guided Transformer for Multivariate Time Series Anomaly Detection

LGAT: A Novel Model for Multivariate Time Series Anomaly Detection with Imporved Anomaly Transformer and Learning Graph Structures

Multivariate Time Series Anomaly Detection Based on Spatial-Temporal Network and Transformer in Industrial Internet of Things

Unsupervised Anomaly Detection in Multivariate Time Series Through Transformer-based Variational Autoencoder

Sub-Adjacent Transformer: Improving Time Series Anomaly Detection with Reconstruction Error from Sub-Adjacent Neighborhoods

TCF-Trans: Temporal Context Fusion Transformer for Anomaly Detection in Time Series

DTAAD: Dual Tcn-Attention Networks for Anomaly Detection in Multivariate Time Series Data

Unsupervised Spatio-Temporal State Estimation for Fine-grained Adaptive Anomaly Diagnosis of Industrial Cyber-physical Systems

SiET: Spatial information enhanced transformer for multivariate time series anomaly detection

Multivariate time series anomaly detection with adversarial transformer architecture in the Internet of Things

Disentangled Dynamic Deviation Transformer Networks for Multivariate Time Series Anomaly Detection

Multivariate Time Series Anomaly Detection Via Separation, Decomposition, and Dual Transformer-Based Autoencoder

EdgeConvFormer: Dynamic Graph CNN and Transformer based Anomaly Detection in Multivariate Time Series