Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Online Model-based Anomaly Detection in Multivariate Time Series: Taxonomy, Survey, Research Challenges and Future Directions

Lucas Correia, Jan-Christoph Goos, Philipp Klein, Thomas Bäck, Anna V. Kononova

TL;DR

This survey provides an extensive overview of the state-of-the-art model-based online semi- and unsupervised anomaly detection approaches for multivariate time-series data, categorising them into different model families and other properties.

Abstract

Time-series anomaly detection plays an important role in engineering processes, like development, manufacturing and other operations involving dynamic systems. These processes can greatly benefit from advances in the field, as state-of-the-art approaches may aid in cases involving, for example, highly dimensional data. To provide the reader with understanding of the terminology, this survey introduces a novel taxonomy where a distinction between online and offline, and training and inference is made. Additionally, it presents the most popular data sets and evaluation metrics used in the literature, as well as a detailed analysis. Furthermore, this survey provides an extensive overview of the state-of-the-art model-based online semi- and unsupervised anomaly detection approaches for multivariate time-series data, categorising them into different model families and other properties. The biggest research challenge revolves around benchmarking, as currently there is no reliable way to compare different approaches against one another. This problem is two-fold: on the one hand, public data sets suffers from at least one fundamental flaw, while on the other hand, there is a lack of intuitive and representative evaluation metrics in the field. Moreover, the way most publications choose a detection threshold disregards real-world conditions, which hinders the application in the real world. To allow for tangible advances in the field, these issues must be addressed in future work.

Online Model-based Anomaly Detection in Multivariate Time Series: Taxonomy, Survey, Research Challenges and Future Directions

TL;DR

This survey provides an extensive overview of the state-of-the-art model-based online semi- and unsupervised anomaly detection approaches for multivariate time-series data, categorising them into different model families and other properties.

Abstract

Time-series anomaly detection plays an important role in engineering processes, like development, manufacturing and other operations involving dynamic systems. These processes can greatly benefit from advances in the field, as state-of-the-art approaches may aid in cases involving, for example, highly dimensional data. To provide the reader with understanding of the terminology, this survey introduces a novel taxonomy where a distinction between online and offline, and training and inference is made. Additionally, it presents the most popular data sets and evaluation metrics used in the literature, as well as a detailed analysis. Furthermore, this survey provides an extensive overview of the state-of-the-art model-based online semi- and unsupervised anomaly detection approaches for multivariate time-series data, categorising them into different model families and other properties. The biggest research challenge revolves around benchmarking, as currently there is no reliable way to compare different approaches against one another. This problem is two-fold: on the one hand, public data sets suffers from at least one fundamental flaw, while on the other hand, there is a lack of intuitive and representative evaluation metrics in the field. Moreover, the way most publications choose a detection threshold disregards real-world conditions, which hinders the application in the real world. To allow for tangible advances in the field, these issues must be addressed in future work.
Paper Structure (31 sections, 16 equations, 11 figures, 9 tables)

This paper contains 31 sections, 16 equations, 11 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Taxonomy in Time-series Anomaly Detection
  3. Anomaly Types and Detection
  4. Continuous- and Discrete-sequence Anomaly Detection
  5. Online Training and Inference
  6. Related Work
  7. Benchmarking
  8. Data Sets
  9. Metrics
  10. Calibrated metrics
  11. Uncalibrated metrics
  12. Predictive Models
  13. Online Training and Online Inference
  14. Offline Training and Online Inference
  15. Offline Training and Offline Inference but Online Capable
  16. ...and 16 more sections

Figures (11)

  • Figure 1: A visual representation of the structure of the survey.
  • Figure 2: A visual representation of taxonomy used in this survey.
  • Figure 3: Example of a point anomaly in both channels. The continuous line represents the ground-truth time series and the traced line the anomaly. The ground-truth time series shown is a heartbeat from the MIT-BIH Arrhythmia Database moody_impact_2001.
  • Figure 4: Example of a sub-sequence anomaly in both channels. The continuous line represents the ground-truth time series and the traced line the anomaly. The ground-truth time series shown is a heartbeat from the MIT-BIH Arrhythmia Database moody_impact_2001.
  • Figure 5: Example of a whole-sequence anomaly in both channels. The continuous line represents the ground-truth time series and the traced line the anomaly. The ground-truth time series shown is a heartbeat from the MIT-BIH Arrhythmia Database moody_impact_2001.
  • ...and 6 more figures