KWKBW Philip - … CVPR IEEE Computer Society Conference on …, 1996 - computer.org
… The simultaneous minimization is either approached as a multi-objectiveproblem or with aggregating functions combining mean andDesign SpaceParameter Min Max -n128 -n64 -n32 -…
… We set MFS's subset size parameter basedon crossvalidation accuracyestimates on the training set for the entire ensemble. We evaluated ten evenly spaced intervals over the size of …
H Cevikalp, R Polikar - IEEE Transactions on Neural Networks, 2008 - ieeexplore.ieee.org
… , we uselocalclassifieraccuracyestimates to weight classifier … , we estimatelocal recognition accuracies of classifiers near a … , “Combination of multiple classifiersusinglocalaccuracy …
BM Steele - Remote sensing of environment, 2000 - Elsevier
… , the product rule, for combiningmultipleclassification rules, and I … regression method of combiningclassification rules. To … accuracyestimates for the three combinationclassifiers and …
S Sun - Pattern Recognition Letters, 2010 - Elsevier
… To begin with, we discuss several related combination … After introducing the local within-class accuracy method, we conduct an … classifier selection by means of localaccuracyestimates. …
FM Campos, L Correia, JMF Calado - Journal of Intelligent & Robotic …, 2015 - Springer
… more accurate approach that relies on continuous valued scores obtained from local features. … of our approach, by describing how posterior estimates are obtained from a local feature. …
PC Smits - IEEE Transactions on Geoscience and Remote …, 2002 - ieeexplore.ieee.org
… are presented that useestimates of each individual classifier’s localaccuracy in small … For more generic reviews on multipleclassifier systems and combiningclassifiers the reader is …
… in determining the overall accuracy of the classifiercombinations. Therefore, to minimize the … It involves the weighting of classifiersbasedon their estimated recognition performance, to …
T Windeatt - IEEE Transactions on Neural Networks, 2006 - ieeexplore.ieee.org
… combiningmultipleclassifiers is basedon the observation that achieving optimal performance in combination … incorporation of bootstrapping for estimating the measures enables good …