On compositions of transformations in contrastive self-supervised learning

In the image domain, excellent representations can be learned by inducing invariance to content-preserving transformations via noise contrastive learning. In this paper, we generalize contrastive learning to a wider set of transformations, and their compositions, for which either invariance or disti...

Бүрэн тодорхойлолт

Номзүйн дэлгэрэнгүй
Үндсэн зохиолчид: Yuki M. Asano, YM, Patrick, M, Kuznetsova, P, Fong, R, Henriques, JF, Zweig, G, Vedaldi, A
Формат: Conference item
Хэл сонгох:English
Хэвлэсэн: IEEE 2022

Ижил төстэй зүйлс