File size: 147 Bytes
5fa1a76
1
The portion of the unlabeled data that the model predicts with the most confidence gets added to the labeled dataset and used to retrain the model.