The paper proposes a theory-based method for estimating the optimal value of k in k-NN classifiers based on a n-sized training set As expected, experiments show that the suggested k is such that k/n →0 when both k and n tend to infinity, as required by the asymptotical consistency condition. Interestingly, it appears that the generalization error is robust w.r.t. to k when n becomes large (probably as a consequence of the k/n→0 relationship); the immediate consequence is that there is no need to provide an accurate estimate for the optimal k and an approximated coarser value, e.g., provided with cross validation, I-fold cross validation or leave one out is more than adequate.
K-NN classifiers: Investigating the k = k(n) relationship / C. Alippi, M. Fuhrman, M. Roveri (PROCEEDINGS OF ... INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS). - In: 2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence)[s.l] : IEEE, 2008. - ISBN 9781424418206. - pp. 3676-3680 (( convegno International Joint Conference on Neural Networks tenutosi a Hong Kong nel 2008.
K-NN classifiers: Investigating the k = k(n) relationship
M. Fuhrman;M. Roveri
2008
Abstract
The paper proposes a theory-based method for estimating the optimal value of k in k-NN classifiers based on a n-sized training set As expected, experiments show that the suggested k is such that k/n →0 when both k and n tend to infinity, as required by the asymptotical consistency condition. Interestingly, it appears that the generalization error is robust w.r.t. to k when n becomes large (probably as a consequence of the k/n→0 relationship); the immediate consequence is that there is no need to provide an accurate estimate for the optimal k and an approximated coarser value, e.g., provided with cross validation, I-fold cross validation or leave one out is more than adequate.File | Dimensione | Formato | |
---|---|---|---|
IJCNN08_Alippi_Fuhrman_Roveri.pdf
accesso riservato
Tipologia:
Post-print, accepted manuscript ecc. (versione accettata dall'editore)
Dimensione
234.88 kB
Formato
Adobe PDF
|
234.88 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
04634324.pdf
accesso riservato
Tipologia:
Publisher's version/PDF
Dimensione
268.19 kB
Formato
Adobe PDF
|
268.19 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.