DyTox: Transformers for Continual Learning with DYnamic TOken eXpansion

Deep network architectures struggle to continually learn new tasks without forgetting the previous tasks. A recent trend indicates that dynamic architectures based on an ex-pansion of the parameters can reduce catastrophic forget-ting efficiently in continual learning. However, existing approaches o...

Celý popis

Podrobná bibliografia

Vydané v:	Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online) s. 9275 - 9285
Hlavní autori:	Douillard, Arthur, Rame, Alexandre, Couairon, Guillaume, Cord, Matthieu
Médium:	Konferenčný príspevok..
Jazyk:	English
Vydavateľské údaje:	IEEE 01.06.2022
Predmet:	Decoding Deep learning Memory management Network architecture Representation learning Transfer/low-shot/long-tail learning; Deep learning architectures and techniques; Machine learning; Representation learning Transformers
ISSN:	1063-6919
On-line prístup:	Získať plný text
Tagy:	Pridať tag Žiadne tagy, Buďte prvý, kto otaguje tento záznam!