# Gaussian Process Latent Variable Models for Fault Detection

Luka Eciolaza
M. Alkarouri
Neil D. Lawrence, University of Sheffield
Peter J. Fleming

in Computational Intelligence and Data Mining, pp 287-292

#### Abstract

The Gaussian process latent variable model (GPLVM) is a novel unsupervised approach to nonlinear low dimensional embedding proposed by Lawrence (2005). This paper presents the development of a framework for the implementation of the GPLVM for fault detection. A series of experiments have been carried out comparing and combining the GPLVM to the conventional and widely used linear dimension reduction technique of principal component analysis (PCA). The inclusion of the GPLVM for the visualisation and data analysis, led to a considerable improvement in the classification results

  @InProceedings{eciolaza-fault07, title = {Gaussian Process Latent Variable Models for Fault Detection}, author = {Luka Eciolaza and M. Alkarouri and Neil D. Lawrence and Visakan Kadirkamanathan and Peter J. Fleming}, booktitle = {Computational Intelligence and Data Mining}, pages = {287}, year = {2007}, month = {00}, edit = {https://github.com/lawrennd//publications/edit/gh-pages/_posts/2007-01-01-eciolaza-fault07.md}, url = {http://inverseprobability.com/publications/eciolaza-fault07.html}, abstract = {The Gaussian process latent variable model (GPLVM) is a novel unsupervised approach to nonlinear low dimensional embedding proposed by Lawrence (2005). This paper presents the development of a framework for the implementation of the GPLVM for fault detection. A series of experiments have been carried out comparing and combining the GPLVM to the conventional and widely used linear dimension reduction technique of principal component analysis (PCA). The inclusion of the GPLVM for the visualisation and data analysis, led to a considerable improvement in the classification results}, key = {Eciolaza:fault07}, doi = {10.1109/CIDM.2007.368886}, group = {gplvm} }
 %T Gaussian Process Latent Variable Models for Fault Detection %A Luka Eciolaza and M. Alkarouri and Neil D. Lawrence and Visakan Kadirkamanathan and Peter J. Fleming %B %C Computational Intelligence and Data Mining %D %F eciolaza-fault07 %P 287--292 %R 10.1109/CIDM.2007.368886 %U http://inverseprobability.com/publications/eciolaza-fault07.html %X The Gaussian process latent variable model (GPLVM) is a novel unsupervised approach to nonlinear low dimensional embedding proposed by Lawrence (2005). This paper presents the development of a framework for the implementation of the GPLVM for fault detection. A series of experiments have been carried out comparing and combining the GPLVM to the conventional and widely used linear dimension reduction technique of principal component analysis (PCA). The inclusion of the GPLVM for the visualisation and data analysis, led to a considerable improvement in the classification results 
 TY - CPAPER TI - Gaussian Process Latent Variable Models for Fault Detection AU - Luka Eciolaza AU - M. Alkarouri AU - Neil D. Lawrence AU - Visakan Kadirkamanathan AU - Peter J. Fleming BT - Computational Intelligence and Data Mining PY - 2007/01/01 DA - 2007/01/01 ID - eciolaza-fault07 SP - 287 EP - 292 DO - 10.1109/CIDM.2007.368886 UR - http://inverseprobability.com/publications/eciolaza-fault07.html AB - The Gaussian process latent variable model (GPLVM) is a novel unsupervised approach to nonlinear low dimensional embedding proposed by Lawrence (2005). This paper presents the development of a framework for the implementation of the GPLVM for fault detection. A series of experiments have been carried out comparing and combining the GPLVM to the conventional and widely used linear dimension reduction technique of principal component analysis (PCA). The inclusion of the GPLVM for the visualisation and data analysis, led to a considerable improvement in the classification results ER - 
 Eciolaza, L., Alkarouri, M., Lawrence, N.D., Kadirkamanathan, V. & Fleming, P.J.. (2007). Gaussian Process Latent Variable Models for Fault Detection. Computational Intelligence and Data Mining :287-292