Confident Sinkhorn Allocation for Pseudo-Labeling

Open in new window