Psuedo-labelling in Semi-Supervised Learning
Psuedo-labelling is a technique where a model is first trained on a small amount of labeled data and then used to generate labels for a larger unlabeled dataset. These predicted labels are then used to further train the model, effectively leveraging unlabeled data to improve accuracy and generalization.