Test-Time Training with Masked Autoencoders Y ossi Gandelsman UC Berkeley Y u Sun

Neural Information Processing Systems 

In this paper, we use masked autoen-coders for this one-sample learning problem.