Fig. 5From: Comparison of semi-supervised deep learning algorithms for audio classificationFixMatch workflow. A weakly augmented version of xu is used to compute a pseudo-label \(\hat {y}_{u}\) and a mask. The strongly augmented variant is used to compute the unlabeled loss term. The mixup component is used on a concatenated set of labeled and unlabeled samples (FixMatch+mixup)Back to article page