Forskning ved Københavns Universitet - Københavns Universitet

Forside

Weighting of the k-Nearest-Neighbors

Publikation: Bidrag til bog/antologi/rapportKonferencebidrag i proceedingsForskningfagfællebedømt

This paper presents two distribution independent weighting schemes for k-Nearest-Neighbors (kNN). Applying the first scheme in a Leave-One-Out (LOO) setting corresponds to performing complete b-fold cross validation (b-CCV), while applying the second scheme corresponds to performing bootstrapping in the limit of infinite iterations. We demonstrate that the soft kNN errors obtained through b-CCV can be obtained by applying the weighted kNN in a LOO setting, and that the proposed weighting schemes can decrease the variance and improve the generalization of kNN in a CV setting.

OriginalsprogEngelsk
Titel2010 20th International Conference on Pattern Recognition (ICPR)
Antal sider4
ForlagIEEE
Publikationsdato2010
Sider666-669
ISBN (Trykt)978-1-4244-7542-1
ISBN (Elektronisk)978-1-4244-7541-4
DOI
StatusUdgivet - 2010
Begivenhed20th International Conference on Pattern Recognition - Istanbul, Tyrkiet
Varighed: 23 aug. 201026 aug. 2010

Konference

Konference20th International Conference on Pattern Recognition
LandTyrkiet
ByIstanbul
Periode23/08/201026/08/2010

ID: 172801044