ReMixMatch: Semi-Supervised Learning with Distribution Alignment and Augmentation Anchoring

note