A maximum uncertainty LDA-based approach for limited sample size problems — with application to face recognition

Thomaz, Carlos Eduardo; Kitani, Edson C.; Gillies, Donald

doi:10.1007/bf03192391

Cited by 39 publications

(23 citation statements)

References 22 publications

(13 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…FLD can be easily extended to multi-class cases via multiple discriminant analysis [41]. In fact, discriminant analysis has been widely used in this way in pattern recognition problems, and its use in reducing facial recognition problems is well documented [42]. This has been the motivation to use discriminant analysis for multiple distinct binary classifications.…”

Section: A the Lda Based Classifiermentioning

confidence: 99%

A Housekeeping Prognostic Health Management Framework for Microfluidic Systems

Khan

Al-Gayem

Richardson

2017

IEEE Trans. Device Mater. Relib.

View full text Add to dashboard Cite

Abstract--Micro-electro-mechanical Systems and microfluidics are becoming popular solutions for sensing, diagnostics and control applications. Reliability and validation of function is of increasing importance in the majority of these applications. Online testing strategies for these devices have the potential to provide real-time condition monitoring information. It is shown that this information can be used to diagnose and prognose the health of the device. This information can also be used to provide an early failure warning system by predicting the remaining useful life. Diagnostic and prognostic outcomes can also be leveraged to improve the reliability, dependability and availability of these devices. This work has delivered a methodology for a "lightweight" prognostics solution for a microfluidic device based on real-time diagnostics. An oscillation based test methodology is used to extract diagnostic information that is processed using a Linear Discriminant Analysis based classifier. This enables the identification of current health based on pre-defined health levels. As the deteriorating device is periodically classified, the rate at which the device degrades is used to predict the devices remaining useful life. I. INTRODUCTIONA novel implementation of a prognostics function for a microfluidic device based on MEMS technology is presented. The term "prognostics" has been defined in various different ways depending on context [1], this research considers the definition given by ISO13381-1 as the most comprehensive. It states that prognostics is not only the prediction of the time to failure but also the risk of one or more existing or future failure modes [2].Studies have revealed some excellent Prognostic and Health Management (PHM) solutions that provide deep and detailed prognostic capabilities for micro-devices with a high level of accuracy. The use of machine learning [3] and physics based prognostics [4] is being actively explored. PHM solutions similar to these are designed from the ground up in a device specific framework with the prognostics modules being tightly coupled with the physics of failure of the device. While these PHM systems perform very well in conjunction with their target devices, flexibility and wide scale adoption in other MEMS devices is quite challenging.The onus of most PHM frameworks is to provide prognosis at the highest possible confidence level thus providing the most reliable Remaining Useful Life (RUL) predictions [5] [6] [7]. However, this comes at a cost, be it a heavy processing requirement or the aforementioned rigidity or complexity, to name two. This work proposes a "Housekeeping" Prognostics and Health Management (HPHM) framework that has a conservative prognostic mandate. Such a framework caters strictly for wear during normal use and evolving faults. This affords the usage of simpler models that are easy to adapt, implement and use.The scope of this HPHM is to predict patterns of low intensity failures and drift due to gradual aging and noncatastrophic faults du...

show abstract

Section: A the Lda Based Classifiermentioning

confidence: 99%

A Housekeeping Prognostic Health Management Framework for Microfluidic Systems

Khan

Al-Gayem

Richardson

2017

IEEE Trans. Device Mater. Relib.

View full text Add to dashboard Cite

show abstract

“…To avoid both critical issues, we have calculated w lda by using a maximum uncertainty LDA-based approach (MLDA) that considers the issue of stabilizing the S w estimate with a multiple of the identity matrix [26,25,27].…”

Section: Linear Discriminant Analysis (Lda)mentioning

confidence: 99%

“…We review the theory behind the cited methods, their common points, and discuss why SVM is in general the best technique for classification but not necessarily the best for extracting discriminant information. This will be discussed using face images and the separating hyperplanes generated by SVM [30,14,28] and a regularized version of LDA called Maximum uncertainty Linear Discriminant Analysis (MLDA) [27]. To further evaluate DFA on a data set not composed of images, we investigate a breast lesion classification framework, proposed in [17], that uses ultrasound features and SVM only.…”

Section: Introductionmentioning

confidence: 99%

Statistical learning approaches for discriminant features selection

et al. 2008

View full text Add to dashboard Cite

“…To avoid the aforementioned critical issues of the standard LDA in limited sample and high dimensional problems, we have calculated lda P by using a maximum uncertainty LDA-based approach (MLDA) that considers the issue of stabilising the w S estimate with a multiple of the identity matrix [8,9]. In a previous study [8] with application to the face recognition problem, Thomaz and Gillies showed that the MLDA approach improved the LDA classification performance with or without a PCA intermediate step and using less linear discriminant features [8].…”

Section: Maximum Uncertainty Lda (Mlda)mentioning

confidence: 99%

“…Analogously to the Cootes et al approaches [1 -4], SDM requires a previous alignment of all the images to a common template to minimise variations that are not necessarily related to differences between the faces. However, instead of using landmarks or annotations on the images, SDM is based on the idea of using PCA to reduce the dimensionality of the original images and a maximum uncertainty linear classifier (MLDA) [8] to characterise the most discriminant differences between the samples of images.…”

Section: Introductionmentioning

confidence: 99%