Hypernetwork approach to Bayesian MAML (Student Abstract)

2025
book section
conference proceedings
dc.abstract.enThe main goal of Few-Shot learning algorithms is to enabl learning from small amounts of data. One of the most popular and elegant Few-Shot learning approaches is Model-Agnostic Meta-Learning (MAML). In this paper, we propose a novel framework for Bayesian MAML called BH-MAML, which employs Hypernetworks for weight updates. It learns the universal weights point-wise, but a probabilistic structure is added when adapted for specific tasks. In such a framework, we can use simple Gaussian distributions or more complicated posteriors induced by Continuous Normalizing Flows
dc.affiliationSzkoła Doktorska Nauk Ścisłych i Przyrodniczych
dc.affiliationWydział Matematyki i Informatyki : Instytut Informatyki i Matematyki Komputerowej
dc.conference39th AAAI Conference on Artificial Intelligence
dc.conference.cityFiladelfia, Pensylwania
dc.conference.countryStany Zjednoczone
dc.conference.datefinish2025-03-04
dc.conference.datestart2025-02-25
dc.conference.seriesNational Conference of the American Association for Artificial Intelligence
dc.conference.seriesshortcutAAAI
dc.conference.shortcutAAAI-25
dc.conference.weblinkhttps://aaai.org/conference/aaai/aaai-25/
dc.contributor.authorBorycki, Piotr
dc.contributor.authorKubacki, Piotr
dc.contributor.authorPrzewięźlikowski, Marcin - 421101
dc.contributor.authorKuśmierczyk, Tomasz - 498199
dc.contributor.authorTabor, Jacek - 132362
dc.contributor.authorSpurek, Przemysław - 135993
dc.contributor.editorWalsh, Toby
dc.contributor.editorShah, Julie
dc.contributor.editorKolter, Zico
dc.date.accessioned2025-05-19T06:33:38Z
dc.date.available2025-05-19T06:33:38Z
dc.date.createdat2025-04-15T07:48:22Zen
dc.date.issued2025
dc.date.openaccess0
dc.description.accesstimew momencie opublikowania
dc.description.conftypeinternational
dc.description.physical29325-29327
dc.description.versionostateczna wersja wydawcy
dc.description.volume39
dc.identifier.bookweblinkhttps://search.worldcat.org/title/10789238955?oclcNum=10789238955
dc.identifier.doi10.1609/aaai.v39i28.35239
dc.identifier.isbn978-1-57735-897-8
dc.identifier.isbn1-57735-897-X
dc.identifier.project2021/41/B/ST6/01370
dc.identifier.project2023/50/E/ST6/00068
dc.identifier.project2023/49/N/ST6/03268
dc.identifier.project2022/45/P/ST6/0296
dc.identifier.projectMarie Sklodowska-Curie grant agreement No. 945339
dc.identifier.serieseissn2374-3468
dc.identifier.seriesissn2159-5399
dc.identifier.urihttps://ruj.uj.edu.pl/handle/item/552534
dc.languageeng
dc.language.containereng
dc.placeWashington
dc.publisherAAAI Press
dc.rightsDodaję tylko opis bibliograficzny
dc.rights.licenceInna otwarta licencja
dc.share.typeinne
dc.source.integratorfalse
dc.subtypeConferenceProceedings
dc.titleHypernetwork approach to Bayesian MAML (Student Abstract)
dc.title.containerProceedings of the 39th AAAI Conference on Artificial Intelligence
dc.title.volumeProceedings of the 39th Annual AAAI Conference on Artificial IntelligenceIAAI-25, EAAI-25, AAAI-25 Student Abstracts, Undergraduate Consortium and Demonstrations
dc.typeBookSection
dspace.entity.typePublicationen
dc.abstract.en
The main goal of Few-Shot learning algorithms is to enabl learning from small amounts of data. One of the most popular and elegant Few-Shot learning approaches is Model-Agnostic Meta-Learning (MAML). In this paper, we propose a novel framework for Bayesian MAML called BH-MAML, which employs Hypernetworks for weight updates. It learns the universal weights point-wise, but a probabilistic structure is added when adapted for specific tasks. In such a framework, we can use simple Gaussian distributions or more complicated posteriors induced by Continuous Normalizing Flows
dc.affiliation
Szkoła Doktorska Nauk Ścisłych i Przyrodniczych
dc.affiliation
Wydział Matematyki i Informatyki : Instytut Informatyki i Matematyki Komputerowej
dc.conference
39th AAAI Conference on Artificial Intelligence
dc.conference.city
Filadelfia, Pensylwania
dc.conference.country
Stany Zjednoczone
dc.conference.datefinish
2025-03-04
dc.conference.datestart
2025-02-25
dc.conference.series
National Conference of the American Association for Artificial Intelligence
dc.conference.seriesshortcut
AAAI
dc.conference.shortcut
AAAI-25
dc.conference.weblink
https://aaai.org/conference/aaai/aaai-25/
dc.contributor.author
Borycki, Piotr
dc.contributor.author
Kubacki, Piotr
dc.contributor.author
Przewięźlikowski, Marcin - 421101
dc.contributor.author
Kuśmierczyk, Tomasz - 498199
dc.contributor.author
Tabor, Jacek - 132362
dc.contributor.author
Spurek, Przemysław - 135993
dc.contributor.editor
Walsh, Toby
dc.contributor.editor
Shah, Julie
dc.contributor.editor
Kolter, Zico
dc.date.accessioned
2025-05-19T06:33:38Z
dc.date.available
2025-05-19T06:33:38Z
dc.date.createdaten
2025-04-15T07:48:22Z
dc.date.issued
2025
dc.date.openaccess
0
dc.description.accesstime
w momencie opublikowania
dc.description.conftype
international
dc.description.physical
29325-29327
dc.description.version
ostateczna wersja wydawcy
dc.description.volume
39
dc.identifier.bookweblink
https://search.worldcat.org/title/10789238955?oclcNum=10789238955
dc.identifier.doi
10.1609/aaai.v39i28.35239
dc.identifier.isbn
978-1-57735-897-8
dc.identifier.isbn
1-57735-897-X
dc.identifier.project
2021/41/B/ST6/01370
dc.identifier.project
2023/50/E/ST6/00068
dc.identifier.project
2023/49/N/ST6/03268
dc.identifier.project
2022/45/P/ST6/0296
dc.identifier.project
Marie Sklodowska-Curie grant agreement No. 945339
dc.identifier.serieseissn
2374-3468
dc.identifier.seriesissn
2159-5399
dc.identifier.uri
https://ruj.uj.edu.pl/handle/item/552534
dc.language
eng
dc.language.container
eng
dc.place
Washington
dc.publisher
AAAI Press
dc.rights
Dodaję tylko opis bibliograficzny
dc.rights.licence
Inna otwarta licencja
dc.share.type
inne
dc.source.integrator
false
dc.subtype
ConferenceProceedings
dc.title
Hypernetwork approach to Bayesian MAML (Student Abstract)
dc.title.container
Proceedings of the 39th AAAI Conference on Artificial Intelligence
dc.title.volume
Proceedings of the 39th Annual AAAI Conference on Artificial IntelligenceIAAI-25, EAAI-25, AAAI-25 Student Abstracts, Undergraduate Consortium and Demonstrations
dc.type
BookSection
dspace.entity.typeen
Publication
Affiliations

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

Views
51
Views per month
Views per city
Krakow
26
Amsterdam
1
Brzesko
1
Singapore
1
Wroclaw
1

No access

No Thumbnail Available