Publicacions CVCDisplays records where serial is equal to 16632024-03-29T15:35:36+01:00Publicacions CVCsi@cvc.uab.eshttp://refbase.cvc.uab.es/Web Reference Database (http://refbase.sourceforge.net)http://refbase.cvc.uab.es/img/favicon.icohttp://refbase.cvc.uab.es/img/logo.pnghttp://refbase.cvc.uab.es/show.php?where=serial%20%3D%201663&showRows=15&citeStyle=Chicago&exportType=xml&submit=Export&exportFormat=Atom%20XML111http://refbase.cvc.uab.es/show.php?record=1663
A Novel FLDA Formulation for Numerical Stability Analysis
Fernando Vilariño, Debora Gil, & Petia Radeva. (2004). “A Novel FLDA Formulation for Numerical Stability Analysis ” In P. R. and I. A. J. Vitrià (Ed.), Recent Advances in Artificial Intelligence Research and Development (Vol. 113, pp. 77–84). IOS Press.
Fisher Linear Discriminant Analysis (FLDA) is one of the most popular techniques used in classification applying dimensional reduction. The numerical scheme involves the inversion of the within-class scatter matrix, which makes FLDA potentially ill-conditioned when it becomes singular. In this paper we present a novel explicit formulation of FLDA in terms of the eccentricity ratio and eigenvector orientations of the within-class scatter matrix. An analysis of this function will characterize those situations where FLDA response is not reliable because of numerical instability. This can solve common situations of poor classification performance in computer vision.
A Novel FLDA Formulation for Numerical Stability AnalysisFernando VilariñoDebora GilPetia RadevaJ. Vitrià, P.R. and I.A.urn:ISBN:978-1-58603-466-5openurl:?ctx_ver=Z39.88-2004&rfr_id=info%3Asid%2Frefbase.cvc.uab.es%2F&genre=bookitem&atitle=A%20Novel%20FLDA%20Formulation%20for%20Numerical%20Stability%20Analysis&title=Recent%20Advances%20in%20Artificial%20Intelligence%20Research%20and%20Development&btitle=Recent%20Advances%20in%20Artificial%20Intelligence%20Research%20and%20Development&isbn=978-1-58603-466-5&date=2004&volume=113&spage=77&epage=84&aulast=Fernando%20Vilari%C3%B1o&au=Debora%20Gil&au=Petia%20Radeva&pub=IOS%20Press&id=http%3A%2F%2Fwww.iospress.nl%2Floadtop%2Fload.php%3Fisbn%3D9781586034665&sid=refbase%3ACVCFernando Vilariño, Debora Gil, & Petia Radeva. (2004).<a target='_blank' href='/files/VGR2004.pdf'> "A Novel FLDA Formulation for Numerical Stability Analysis "</a> In P. R. and I. A. J. Vitrià (Ed.), Recent Advances in Artificial Intelligence Research and Development (Vol. 113, pp. 77-84). IOS Press.2004BookChaptertextSupervised LearningLinear Discriminant AnalysisNumerical StabilityComputer Visionurl:http://www.iospress.nl/loadtop/load.php?isbn=9781586034665file:http://refbase.cvc.uab.es/files/VGR2004.pdfIOS PressRecent Advances in Artificial Intelligence Research and Development20041137784