Faculté des Sciences
Faculté des Sciences

Long Short-Term Memory neural networks and Support Vector Data Description for anomaly detection

Keydener, Jimmy ULiège
Promotor(s) : Heuchenne, Cédric ULiège
Date of defense : 29-Jun-2020/30-Jun-2020 • Permalink :
Title : Long Short-Term Memory neural networks and Support Vector Data Description for anomaly detection
Translated title : [fr] Réseaux de neurones Long Short-Term Memory et Support Vector Data Description appliqués à la détection d'anomalies
Author : Keydener, Jimmy ULiège
Date of defense  : 29-Jun-2020/30-Jun-2020
Advisor(s) : Heuchenne, Cédric ULiège
Committee's member(s) : Charlier, Emilie ULiège
Haesbroeck, Gentiane ULiège
Aboubacar, Amir ULiège
Language : English
Number of pages : 65
Keywords : [en] Long short term memory networks
[en] Anomaly detection
[en] Support vector data description
[en] LSTM
[en] SVDD
[en] Neural networks
[en] Vanishing gradient problem
[en] Auto-Encoder-Decoder
Discipline(s) : Physical, chemical, mathematical & earth Sciences > Mathematics
Target public : Researchers
Professionals of domain
Institution(s) : Université de Liège, Liège, Belgique
Degree: Master en sciences mathématiques, à finalité spécialisée en statistique
Faculty: Master thesis of the Faculté des Sciences


[en] Anomaly detection refers to the problem of finding rare patterns in data which raise suspicions because they do not comply with an expected behavior. We can consider different kinds of applications like intrusion detection, image processing, system health monitoring and sensor networks. For example, an anomalous pattern coming from sensors on a machine could mean that the machine is ready to break.
Most of the current studies on anomaly detection do not consider recent/past events to detect possible new incoming outliers. The use of Long Short-Term Memory (LSTM) networks is then proposed to deal with time dependent data related with anomaly detection problems.
The goal of Support Vector Data Description (SVDD) is to describe a realistic domain for the data, excluding superfluous space. The resulting boundary can then be used to detect outliers.

In this master thesis, we consider a LSTM-based prediction model for sensor readings coming from a pulp and paper manufacturing machine. Anomalies will then result from too large prediction errors. We compare the SVDD and a discrimination rule based on the assumption of normality for the errors. In the final chapter, we show that for a real world applications the Gaussian distribution for the errors cannot hold and that the need of a non-parametric data descriptions using kernels is real.



Access Memoire_Jimmy_Keydener.pdf
Size: 1.34 MB
Format: Adobe PDF


Access Memoire_Code_Jimmy_Keydener.R
Size: 200.16 kB
Format: Unknown


  • Keydener, Jimmy ULiège Université de Liège > Master sc. math., à fin.


Committee's member(s)

  • Charlier, Emilie ULiège Université de Liège - ULiège > Département de mathématique > Mathématiques discrètes
    ORBi View his publications on ORBi
  • Haesbroeck, Gentiane ULiège Université de Liège - ULiège > Département de mathématique > Statistique mathématique
    ORBi View his publications on ORBi
  • Aboubacar, Amir ULiège Université de Liège - ULiège > Département de mathématique > Probabilités et statistique mathématique
    ORBi View his publications on ORBi
  • Total number of views 116
  • Total number of downloads 307

All documents available on MatheO are protected by copyright and subject to the usual rules for fair use.
The University of Liège does not guarantee the scientific quality of these students' works or the accuracy of all the information they contain.