Statistical tests using hinge/ε-sensitive loss
Yükleniyor...
Dosyalar
Tarih
2013
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Springer-Verlag
Erişim Hakkı
info:eu-repo/semantics/closedAccess
Özet
Statistical tests used in the literature to compare algorithms use the misclassification error which is based on the 0/1 loss and square loss for regression. Kernel-based, support vector machine classifiers (regressors) however are trained to minimize the hinge (ε-sensitive) loss and hence they should not be assessed or compared in terms of the 0/1 (square loss) but with the loss measure they are trained to minimize. We discuss how the paired t test can use the hinge (ε-sensitive) loss and show in our experiments that doing that, we can detect differences that the test on error cannot detect, indicating higher power in distinguishing between the behavior of kernel-based classifiers (regressors). Such tests can be generalized to compare L > 2 algorithms.
Açıklama
Anahtar Kelimeler
Algorithms, Decision trees, Information science, Kernel based classifiers, Misclassification error, Neural networks, Statistical tests, Support vector machine classifiers, Support vector machines, T-tests
Kaynak
Computer and Information Sciences III - 27th International Symposium on Computer and Information Sciences, ISCIS 2012
WoS Q Değeri
Scopus Q Değeri
N/A
Cilt
Sayı
Künye
Yıldız, O. T. & Alpaydın, A. İ. E. (2013). Statistical tests using hinge/ε-sensitive loss. Paper presented at the Computer and Information Sciences III - 27th International Symposium on Computer and Information Sciences, ISCIS 2012, 153-160. doi:10.1007/978-1-4471-4594-3-16