Doorgaan naar hoofdnavigatie Doorgaan naar zoeken Ga verder naar hoofdinhoud

Equivariant Representation Learning in the Presence of Stabilizers

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

4 Downloads (Pure)

Samenvatting

We introduce Equivariant Isomorphic Networks (EquIN) – a method for learning representations that are equivariant with respect to general group actions over data. Differently from existing equivariant representation learners, EquIN is suitable for group actions that are not free, i.e., that stabilize data via nontrivial symmetries. EquIN is theoretically grounded in the orbit-stabilizer theorem from group theory. This guarantees that an ideal learner infers isomorphic representations while trained on equivariance alone and thus fully extracts the geometric structure of data. We provide an empirical investigation on image datasets with rotational symmetries and show that taking stabilizers into account improves the quality of the representations.
Originele taal-2Engels
TitelMachine Learning and Knowledge Discovery in Databases: Research Track
SubtitelEuropean Conference, ECML PKDD 2023, Turin, Italy, September 18–22, 2023, Proceedings
RedacteurenDanai Koutra, Claudia Plant, Manuel Gomez Rodriguez, Elena Baralis, Francesco Bonchi
Plaats van productieCham
UitgeverijSpringer
Pagina's693-708
Aantal pagina's16
VolumeIV
ISBN van elektronische versie978-3-031-43421-1
ISBN van geprinte versie978-3-031-43420-4
DOI's
StatusGepubliceerd - 18 sep. 2023

Publicatie series

NaamLecture Notes in Computer Science (LNCS)
Volume14172
ISSN van geprinte versie0302-9743
ISSN van elektronische versie1611-3349
NaamLecture Notes in Artificial Intelligence (LNAI)
Volume14172
ISSN van geprinte versie2945-9133
ISSN van elektronische versie2945-9141

Financiering

Acknowledgements. This work was supported by the Swedish Research Council, the Knut and Alice Wallenberg Foundation and the European Research Council (ERC-BIRD-884807). This work has also received funding from the NWO-TTW Programme “Efficient Deep Learning” (EDL) P16-25.

FinanciersFinanciernummer
Netherlands Organisation for Applied Scientific Research - TNOP16-25
European Union’s Horizon Europe research and innovation programmeERC-BIRD-884807
Vetenskapsrådet

    Vingerafdruk

    Duik in de onderzoeksthema's van 'Equivariant Representation Learning in the Presence of Stabilizers'. Samen vormen ze een unieke vingerafdruk.

    Citeer dit