Semi-Supervised Learning

Semi-supervised learning trains a model using a small amount of labeled data together with a larger amount of unlabeled data. It matters because model performance depends on how data, training, validation, and evaluation are designed, not only on which algorithm is selected.