The peaking phenomenon in semi-supervised learning

Jesse H. Krijthe, Marco Loog

2 Citationer (Scopus)

Abstract

For the supervised least squares classifier, when the number of training objects is smaller than the dimensionality of the data, adding more data to the training set may first increase the error rate before decreasing it. This, possibly counterintuitive, phenomenon is known as peaking. In this work, we observe that a similar but more pronounced version of this phenomenon also occurs in the semi-supervised setting, where instead of labeled objects, unlabeled objects are added to the training set. We explain why the learning curve has a more steep incline and a more gradual decline in this setting through simulation studies and by applying an approximation of the learning curve based on the work by Raudys and Duin.

OriginalsprogEngelsk
Titel Structural, Syntactic, and Statistical Pattern Recognition : Joint IAPR International Workshop, S+SSPR 2016 Mérida, Mexico, November 29 – December 2, 2016 Proceedings
RedaktørerAntonio Robles-Kelly, Marco Loog, Battista Biggio, Francisco Escolano, Richard Wilson
ForlagSpringer
Publikationsdato5 nov. 2016
Sider299-309
Kapitel11
ISBN (Trykt)978-3-319-49054-0
ISBN (Elektronisk)978-3-319-49055-7
DOI
StatusUdgivet - 5 nov. 2016
BegivenhedJoint IAPR International Workshop on Structural Syntactic, and Statistical Pattern Recognition - Mérida, Mexico
Varighed: 29 nov. 20162 dec. 2016

Workshop

WorkshopJoint IAPR International Workshop on Structural Syntactic, and Statistical Pattern Recognition
Land/OmrådeMexico
ByMérida
Periode29/11/201602/12/2016
NavnLecture notes in computer science
Vol/bind10029
ISSN0302-9743

Fingeraftryk

Dyk ned i forskningsemnerne om 'The peaking phenomenon in semi-supervised learning'. Sammen danner de et unikt fingeraftryk.

Citationsformater