Weakly Augmented Variational Autoencoder in Time Series Anomaly Detection

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Wu, Zhangkai, Cao, Longbing, Zhang, Qi, Zhou, Junxian, Chen, Hui
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866929201198661632
author Wu, Zhangkai
Cao, Longbing
Zhang, Qi
Zhou, Junxian
Chen, Hui
author_facet Wu, Zhangkai
Cao, Longbing
Zhang, Qi
Zhou, Junxian
Chen, Hui
contents Due to their unsupervised training and uncertainty estimation, deep Variational Autoencoders (VAEs) have become powerful tools for reconstruction-based Time Series Anomaly Detection (TSAD). Existing VAE-based TSAD methods, either statistical or deep, tune meta-priors to estimate the likelihood probability for effectively capturing spatiotemporal dependencies in the data. However, these methods confront the challenge of inherent data scarcity, which is often the case in anomaly detection tasks. Such scarcity easily leads to latent holes, discontinuous regions in latent space, resulting in non-robust reconstructions on these discontinuous spaces. We propose a novel generative framework that combines VAEs with self-supervised learning (SSL) to address this issue.
format Preprint
id arxiv_https___arxiv_org_abs_2401_03341
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Weakly Augmented Variational Autoencoder in Time Series Anomaly Detection
Wu, Zhangkai
Cao, Longbing
Zhang, Qi
Zhou, Junxian
Chen, Hui
Machine Learning
Due to their unsupervised training and uncertainty estimation, deep Variational Autoencoders (VAEs) have become powerful tools for reconstruction-based Time Series Anomaly Detection (TSAD). Existing VAE-based TSAD methods, either statistical or deep, tune meta-priors to estimate the likelihood probability for effectively capturing spatiotemporal dependencies in the data. However, these methods confront the challenge of inherent data scarcity, which is often the case in anomaly detection tasks. Such scarcity easily leads to latent holes, discontinuous regions in latent space, resulting in non-robust reconstructions on these discontinuous spaces. We propose a novel generative framework that combines VAEs with self-supervised learning (SSL) to address this issue.
title Weakly Augmented Variational Autoencoder in Time Series Anomaly Detection
topic Machine Learning
url https://arxiv.org/abs/2401.03341