A

A Harol

View Pure Profile

Authored

7 records found

Augmented embedding of dissimilarity data into (pseudo-)Euclidean spaces
Pairwiseproximitiesdescribethepropertiesofobjectsintermsoftheirsimilarities.Byusingdi¿erentdistance-basedfunctionsonemayencodedi¿erentcharacteristicsofagivenproblem.However,tousetheframeworkofstatisticalpatternrecognitionsomevectorrepresentationshouldbeconstructed.Oneofthesimples ...

Non-Euclidean or non-metric measures can be informative
StatisticallearningalgorithmsoftenrelyontheEuclideandistance.Inpractice,non-Euclideanornon-metricdissimilaritymeasuresmayarisewhencontours,spectraorshapesarecomparedbyeditdistancesorasaconsequenceofrobustobjectmatching[1,2].Itisanopenissuewhethersuchmeasuresareadvantageousforstat ...

A new model for prototype selection on distance data
Conference paper - A Harol, R.P.W. Duin, M.J.T. Reinders

On euclidean correction for non-euclidean dissimilarities

Pairwise feature evaluation for constructing reduced representations
Journal article - A Harol, C. Lai, E.M. Pekalska, R.P.W. Duin

Augmented embedding of dissimilarity data into (pseudo-) Euclidean spaces

Pairwise selection of features and prototypes
Conference paper - E.M. Pekalska, A Harol, R.P.W. Duin