BYOL works even without batch statistics

Pierre H. Richemond
Pierre H. Richemond
Florent Altché
Florent Altché
Andrew Brock
Andrew Brock
Samuel Smith
Samuel Smith
Razvan Pascanu
Razvan Pascanu
Cited by: 0|Bibtex|Views8|Links

Abstract:

Bootstrap Your Own Latent (BYOL) is a self-supervised learning approach for image representation. From an augmented view of an image, BYOL trains an online network to predict a target network representation of a different augmented view of the same image. Unlike contrastive methods, BYOL does not explicitly use a repulsion term built fr...More

Code:

Data:

Your rating :
0

 

Tags
Comments