Combining accuracy and prior sensitivity for classifier design under prior uncertainty

None, None; None, None

Combining accuracy and prior sensitivity for classifier design under prior uncertainty

Conference Paper (2006)

Author(s)

TCW Landgrebe (TU Delft - Multimedia Computing)

Bob Duin (TU Delft - Multimedia Computing)

Multimedia Computing

CWTS 0.75 <= JFIS < 2.00 Conference contrib. refereed

To reference this document use:

https://resolver.tudelft.nl/uuid:b691487f-30ce-4768-9c44-fc14d2ac0012

More Info

expand_more

Publication Year

2006

Multimedia Computing

Pages (from-to)

512-521

ISBN (print)

3-540-37236-9

Abstract

Consideringtheclassi¿cationprobleminwhichclasspriorsormisallocationcostsarenotknownprecisely,receiveroperatorcharacteristic(ROC)analysishasbecomeastandardtoolinpatternrecognitionforobtainingintegratedperformancemeasurestocopewiththeuncertainty.Similarly,insituationsinwhichpriorsmayvaryinapplication,theROCcanbeusedtoinspectperformanceovertheexpectedrangeofvariation.InthispaperwearguethateventhoughmeasuressuchastheareaundertheROC(AUC)areusefulinobtaininganintegratedperformancemeasureindependentofthepriors,itmayalsobeimportanttoincorporatethesensitivityacrosstheexpectedprior-range.Weshowthataclassi¿ermayresultinagoodAUCscore,butapoor(large)priorsensitivity,whichmaybeundesirable.Amethodologyisproposedthatcombinesbothaccuracyandsensitivity,providinganewmodelselectioncriterionthatisrelevanttocertainproblems.Experimentsshowthatincorporatingsensitivityisveryimportantinsomerealisticscenarios,leadingtobettermodelselectioninsomecases

No files available

Metadata only record. There are no files for this record.