Exploring the influence of the choice of prior of the Variational Auto-Encoder on cybersecurity anomaly detection

Tengfei Yang; Yuansong Qiao; Brian Lee

doi:10.1145/3664476.3670923

Exploring the influence of the choice of prior of the Variational Auto-Encoder on cybersecurity anomaly detection

Tengfei Yang, Yuansong Qiao, Brian Lee

Research and External Services: Midlands

Limerick Institute of Technology,

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

The Variational Auto-Encoder (VAE) is a popular generative model as the variance inference in the latent layer, the prior is an important element to improve inference efficient. This research explored the prior in the VAE by comparing the Normal family distributions and other location-scale family distributions in three aspects (performance, robustness, and complexity) in order to find a suitable prior for cybersecurity anomaly detection. Suitable distributions can improve the detection performance, which was verified at UNSW-NB15 and CIC-IDS-2017.

Original language	English
Title of host publication	ARES 2024 - 19th International Conference on Availability, Reliability and Security, Proceedings
Publisher	Association for Computing Machinery
ISBN (Electronic)	9798400717185
DOIs	https://doi.org/10.1145/3664476.3670923
Publication status	Published - 30 Jul 2024
Event	19th International Conference on Availability, Reliability and Security, ARES 2024 - Vienna, Austria Duration: 30 Jul 2024 → 2 Aug 2024

Publication series

Name	ACM International Conference Proceeding Series

Conference

Conference	19th International Conference on Availability, Reliability and Security, ARES 2024
Country/Territory	Austria
City	Vienna
Period	30/07/24 → 2/08/24

Keywords

Cybersecurity anomaly detection
Latent representation
Location-scale Distribution
Normal family Distribution
Prior distribution
Variational Auto-Encoder

Access to Document

10.1145/3664476.3670923

Cite this

Yang, T., Qiao, Y., & Lee, B. (2024). Exploring the influence of the choice of prior of the Variational Auto-Encoder on cybersecurity anomaly detection. In ARES 2024 - 19th International Conference on Availability, Reliability and Security, Proceedings Article 67 (ACM International Conference Proceeding Series). Association for Computing Machinery. https://doi.org/10.1145/3664476.3670923

@inproceedings{1b929a2d8341461e9964240661ec4ca5,

title = "Exploring the influence of the choice of prior of the Variational Auto-Encoder on cybersecurity anomaly detection",

abstract = "The Variational Auto-Encoder (VAE) is a popular generative model as the variance inference in the latent layer, the prior is an important element to improve inference efficient. This research explored the prior in the VAE by comparing the Normal family distributions and other location-scale family distributions in three aspects (performance, robustness, and complexity) in order to find a suitable prior for cybersecurity anomaly detection. Suitable distributions can improve the detection performance, which was verified at UNSW-NB15 and CIC-IDS-2017.",

keywords = "Cybersecurity anomaly detection, Latent representation, Location-scale Distribution, Normal family Distribution, Prior distribution, Variational Auto-Encoder",

author = "Tengfei Yang and Yuansong Qiao and Brian Lee",

note = "Publisher Copyright: {\textcopyright} 2024 ACM.; 19th International Conference on Availability, Reliability and Security, ARES 2024 ; Conference date: 30-07-2024 Through 02-08-2024",

year = "2024",

month = jul,

day = "30",

doi = "10.1145/3664476.3670923",

language = "English",

series = "ACM International Conference Proceeding Series",

publisher = "Association for Computing Machinery",

booktitle = "ARES 2024 - 19th International Conference on Availability, Reliability and Security, Proceedings",

address = "United States",

}

Yang, T, Qiao, Y & Lee, B 2024, Exploring the influence of the choice of prior of the Variational Auto-Encoder on cybersecurity anomaly detection. in ARES 2024 - 19th International Conference on Availability, Reliability and Security, Proceedings., 67, ACM International Conference Proceeding Series, Association for Computing Machinery, 19th International Conference on Availability, Reliability and Security, ARES 2024, Vienna, Austria, 30/07/24. https://doi.org/10.1145/3664476.3670923

Exploring the influence of the choice of prior of the Variational Auto-Encoder on cybersecurity anomaly detection. / Yang, Tengfei; Qiao, Yuansong ; Lee, Brian.
ARES 2024 - 19th International Conference on Availability, Reliability and Security, Proceedings. Association for Computing Machinery, 2024. 67 (ACM International Conference Proceeding Series).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Exploring the influence of the choice of prior of the Variational Auto-Encoder on cybersecurity anomaly detection

AU - Yang, Tengfei

AU - Qiao, Yuansong

AU - Lee, Brian

PY - 2024/7/30

Y1 - 2024/7/30

N2 - The Variational Auto-Encoder (VAE) is a popular generative model as the variance inference in the latent layer, the prior is an important element to improve inference efficient. This research explored the prior in the VAE by comparing the Normal family distributions and other location-scale family distributions in three aspects (performance, robustness, and complexity) in order to find a suitable prior for cybersecurity anomaly detection. Suitable distributions can improve the detection performance, which was verified at UNSW-NB15 and CIC-IDS-2017.

AB - The Variational Auto-Encoder (VAE) is a popular generative model as the variance inference in the latent layer, the prior is an important element to improve inference efficient. This research explored the prior in the VAE by comparing the Normal family distributions and other location-scale family distributions in three aspects (performance, robustness, and complexity) in order to find a suitable prior for cybersecurity anomaly detection. Suitable distributions can improve the detection performance, which was verified at UNSW-NB15 and CIC-IDS-2017.

KW - Cybersecurity anomaly detection

KW - Latent representation

KW - Location-scale Distribution

KW - Normal family Distribution

KW - Prior distribution

KW - Variational Auto-Encoder

UR - http://www.scopus.com/inward/record.url?scp=85200381348&partnerID=8YFLogxK

U2 - 10.1145/3664476.3670923

DO - 10.1145/3664476.3670923

M3 - Conference contribution

AN - SCOPUS:85200381348

T3 - ACM International Conference Proceeding Series

BT - ARES 2024 - 19th International Conference on Availability, Reliability and Security, Proceedings

PB - Association for Computing Machinery

T2 - 19th International Conference on Availability, Reliability and Security, ARES 2024

Y2 - 30 July 2024 through 2 August 2024

ER -

Exploring the influence of the choice of prior of the Variational Auto-Encoder on cybersecurity anomaly detection

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this