Systems monitoring based on dynamic classification with SVDD

Published 1 March 2013

Foued Theljani, Kaouther Laabidi, Salah Zidi, Moufida Ksouri

DOI10.1109/ssd.2013.6564010

Citations2

TL;DR

This paper proposes a dynamic classification method based on modified version of the Support Vector Domain Description (SVDD) that supports evolving data and maintains dynamically the descriptive model and is applied on a hydraulic system consisting of three interconnected tanks.

Abstract

In this paper, we address the problem of system monitoring and faults detection using classification-based approach. The main is to follow online evolutions which can occur on the diagnosed system in the course of time. In data classification, the functioning modes are represented with a set of similar patterns called classes. These classes change their intrinsic characteristics and they are likely to be dynamic. Indeed, over the time, new data may be incorporated into the informative model and some others can be likewise discarded. In this paradigm, we propose a dynamic classification method based on modified version of the Support Vector Domain Description (SVDD). Thanks to added insertion/removal procedure, the employed SVDD supports evolving data and maintains dynamically the descriptive model. The proposed approach is applied afterwards on a hydraulic system consisting of three interconnected tanks.

Keywords

Computer ScienceEngineering

TechnometricsStatistical Learning Theory

26,913 Citations1999Yuhai Wu, Vladimir Vapnik

Presenting a method for determining the necessary and sufficient conditions for consistency of learning process, the author covers function estimates from small data pools, applying these estimations to real-life problems, and much more.

Machine LearningSupport Vector Data Description

3,484 Citations2003David M. J. Tax, Robert P. W. Duin

The Support Vector Data Description (SVDD) is presented which obtains a spherically shaped boundary around a dataset and analogous to the Support Vector Classifier it can be made flexible by using other kernel functions.

View PDF

Pattern Recognition LettersSupport vector domain description

1,593 Citations1999David M. J. Tax, Robert P. W. Duin

This paper shows the use of a data domain description method, inspired by the support vector machine by Vapnik, called the support vectors domain description (SVDD), which can be used for novelty or outlier detection and is compared with other outlier Detection methods on real data.

An improved training algorithm for support vector machines

1,066 Citations2002E. Osuna, Robert M. Freund +1 more

This paper presents a decomposition algorithm that is guaranteed to solve the QP problem and that does not make assumptions on the expected number of support vectors.

Lecture notes in computer scienceIncremental One-Class Learning with Bounded Computational Complexity

111 Citations2007Rowland R. Sillito, Robert B. Fisher

An incremental one-class learning algorithm is proposed for the purpose of outlier detection that is shown to outperform a current state-of-the-art incremental one-class learning algorithm (Incremental SVDD [5]) on a variety of datasets, while requiring only an upper limit on model complexity to be specified.

Pattern RecognitionGeneral support vector representation machine for one-class classification of non-stationary classes

55 Citations2008Fatih Camci, Ratna Babu Chinnam

This paper proposes a one-class classification method for non-stationary classes using a modified support vector machine and an efficient online version for reducing computational time.

Journal of SoftwareIncremental Learning Algorithm for Support Vector Data Description

31 Citations2011Xiaopeng Hua, Shifei Ding

A novel algorithm for SVDD incremental learning is proposed, in which the useless sample is discarded and useful information in training samples is accumulated and results indicate the effectiveness of the proposed algorithm.

Constructing a Fast Algorithm for Multi-label Classification with Support Vector Data Description

12 Citations2010Jianhua Xu

One-by-one data decomposition trick is adopted to divide a k-label problem into k sub-problems, where a specific sub-problem only consists of instances with a specific class.

Convex hull based clustering algorithm

12 Citations2013Foued Theljani, Kaouther Laabidi +2 more

A dynamic convex hull based clustering algorithm dealing with data appearing sequentially and an additional merge mechanism is proposed to avoid local optima drawbacks and improve performances.