Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Learning Advancements in Anomaly Detection: A Comprehensive Survey

Haoqi Huang, Ping Wang, Jianhua Pei, Jiacheng Wang, Shahen Alexanian, Dusit Niyato

TL;DR

This survey addresses how deep learning has advanced anomaly detection by organizing methods into reconstruction-, prediction-, and hybrid-based approaches, and by examining how traditional methods can be integrated to improve interpretability and reliability. It aggregates insights from over 180 studies published between 2019 and 2024, detailing GAN-, VAE-, diffusion-, Transformer-, and GNN-based AD developments across time-series, non-temporal, and visual data modalities. Key contributions include structured comparisons, tables of strengths and weaknesses, and a discussion of open issues such as data collection, computational complexity, and explainability, with practical guidance for deploying robust AD systems. The work underscores the growing importance of hybrid models that combine DL flexibility with traditional interpretability to meet real-world demands in IoT, cybersecurity, healthcare, and finance.

Abstract

The rapid expansion of data from diverse sources has made anomaly detection (AD) increasingly essential for identifying unexpected observations that may signal system failures, security breaches, or fraud. As datasets become more complex and high-dimensional, traditional detection methods struggle to effectively capture intricate patterns. Advances in deep learning have made AD methods more powerful and adaptable, improving their ability to handle high-dimensional and unstructured data. This survey provides a comprehensive review of over 180 recent studies, focusing on deep learning-based AD techniques. We categorize and analyze these methods into reconstruction-based and prediction-based approaches, highlighting their effectiveness in modeling complex data distributions. Additionally, we explore the integration of traditional and deep learning methods, highlighting how hybrid approaches combine the interpretability of traditional techniques with the flexibility of deep learning to enhance detection accuracy and model transparency. Finally, we identify open issues and propose future research directions to advance the field of AD. This review bridges gaps in existing literature and serves as a valuable resource for researchers and practitioners seeking to enhance AD techniques using deep learning.

Deep Learning Advancements in Anomaly Detection: A Comprehensive Survey

TL;DR

This survey addresses how deep learning has advanced anomaly detection by organizing methods into reconstruction-, prediction-, and hybrid-based approaches, and by examining how traditional methods can be integrated to improve interpretability and reliability. It aggregates insights from over 180 studies published between 2019 and 2024, detailing GAN-, VAE-, diffusion-, Transformer-, and GNN-based AD developments across time-series, non-temporal, and visual data modalities. Key contributions include structured comparisons, tables of strengths and weaknesses, and a discussion of open issues such as data collection, computational complexity, and explainability, with practical guidance for deploying robust AD systems. The work underscores the growing importance of hybrid models that combine DL flexibility with traditional interpretability to meet real-world demands in IoT, cybersecurity, healthcare, and finance.

Abstract

The rapid expansion of data from diverse sources has made anomaly detection (AD) increasingly essential for identifying unexpected observations that may signal system failures, security breaches, or fraud. As datasets become more complex and high-dimensional, traditional detection methods struggle to effectively capture intricate patterns. Advances in deep learning have made AD methods more powerful and adaptable, improving their ability to handle high-dimensional and unstructured data. This survey provides a comprehensive review of over 180 recent studies, focusing on deep learning-based AD techniques. We categorize and analyze these methods into reconstruction-based and prediction-based approaches, highlighting their effectiveness in modeling complex data distributions. Additionally, we explore the integration of traditional and deep learning methods, highlighting how hybrid approaches combine the interpretability of traditional techniques with the flexibility of deep learning to enhance detection accuracy and model transparency. Finally, we identify open issues and propose future research directions to advance the field of AD. This review bridges gaps in existing literature and serves as a valuable resource for researchers and practitioners seeking to enhance AD techniques using deep learning.

Paper Structure

This paper contains 48 sections, 11 equations, 7 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Contrasting Traditional Models with Deep Learning Models
  3. Comparison With Existing Surveys
  4. Contributions and Structure
  5. Data Characteristics and Challenges
  6. Overview of Input Data and Anomaly Types
  7. Categorization by Data Type
  8. Categorization by Temporal Characteristics
  9. Types of Anomalies
  10. Data Processing
  11. Dimensionality
  12. Sparsity
  13. Class Imbalance
  14. Noise in Data
  15. Privacy of data
  16. ...and 33 more sections

Figures (7)

  • Figure 1: The anatomy of this survey.
  • Figure 2: Three types of anomaly detection: (a) Reconstruction-based approache, (b) Prediction-based approache, (c) Hybrid method.
  • Figure 3: Structural Frameworks for GAN Anomaly Detection.
  • Figure 4: Structural Frameworks for VAE Anomaly Detection.
  • Figure 5: RNN-based application example for time series data anomaly detection: (a) RNN-based, (b) LSTM-based, (c) GRU-based.
  • ...and 2 more figures