Tasks

View-invariance learning

View-invariant models produce similar representations for different transformations ("views") of the same source image.
Using rotation as a transformation example - a view-invariant model would produce similar representations for differently rotated versions of the same image, which could be a useful property for downstream tasks in the context of transfer learning.
This is a self-supervised task.

Dataset format

Datasets follow this structure:

endpoint_url/bucket
└── prefix/images/