The general framework for few-shot learning by kernel HyperNetworks

2023
journal article
article
cris.lastimport.scopus2024-04-07T16:35:35Z
dc.abstract.enFew-shot models aim at making predictions using a minimal number of labeled examples from a given task. The main challenge in this area is the one-shot setting, where only one element represents each class. We propose the general framework for few-shot learning via kernel HyperNetworks—the fusion of kernels and hypernetwork paradigm. Firstly, we introduce the classical realization of this framework, dubbed HyperShot. Compared to reference approaches that apply a gradient-based adjustment of the parameters, our models aim to switch the classification module parameters depending on the task’s embedding. In practice, we utilize a hypernetwork, which takes the aggregated information from support data and returns the classifier’s parameters handcrafted for the considered problem. Moreover, we introduce the kernel-based representation of the support examples delivered to hypernetwork to create the parameters of the classification module. Consequently, we rely on relations between the support examples’ embeddings instead of the backbone models’ direct feature values. Thanks to this approach, our model can adapt to highly different tasks. While such a method obtains very good results, it is limited by typical problems such as poorly quantified uncertainty due to limited data size. We further show that incorporating Bayesian neural networks into our general framework, an approach we call BayesHyperShot, solves this issue.pl
dc.affiliationWydział Matematyki i Informatyki : Instytut Informatyki i Matematyki Komputerowejpl
dc.affiliationSzkoła Doktorska Nauk Ścisłych i Przyrodniczychpl
dc.contributor.authorSendera, Marcin - 246802 pl
dc.contributor.authorPrzewięźlikowski, Marcin - 421101 pl
dc.contributor.authorMiksa, Janpl
dc.contributor.authorRajski, Mateuszpl
dc.contributor.authorKaranowski, Konradpl
dc.contributor.authorZięba, Maciejpl
dc.contributor.authorTabor, Jacek - 132362 pl
dc.contributor.authorSpurek, Przemysław - 135993 pl
dc.date.accessioned2023-06-19T12:16:28Z
dc.date.available2023-06-19T12:16:28Z
dc.date.issued2023pl
dc.date.openaccess0
dc.description.accesstimew momencie opublikowania
dc.description.versionostateczna wersja wydawcy
dc.description.volume34pl
dc.identifier.articleid53pl
dc.identifier.doi10.1007/s00138-023-01403-4pl
dc.identifier.eissn1432-1769pl
dc.identifier.issn0932-8092pl
dc.identifier.urihttps://ruj.uj.edu.pl/xmlui/handle/item/312417
dc.languageengpl
dc.language.containerengpl
dc.rightsUdzielam licencji. Uznanie autorstwa 4.0 Międzynarodowa*
dc.rights.licenceCC-BY
dc.rights.urihttp://creativecommons.org/licenses/by/4.0/legalcode.pl*
dc.share.typeinne
dc.subject.enfew-shot learningpl
dc.subject.enmeta-learningpl
dc.subject.enHyperNetworkspl
dc.subject.enKernel methodspl
dc.subject.enBayesian neural networkspl
dc.subtypeArticlepl
dc.titleThe general framework for few-shot learning by kernel HyperNetworkspl
dc.title.journalMachine Vision and Applicationspl
dc.typeJournalArticlepl
dspace.entity.typePublication

* The migration of download and view statistics prior to the date of April 8, 2024 is in progress.

Views
0
Views per month
Downloads
sendera_et-al_the_general_framework_for_few_shot_learning_2023.pdf
2