What is active learning?

Active learning is an approach in machine learning where the algorithm itself selects the most informative examples for labeling and training. Instead of training on a large volume of randomly labeled data, the model requests labels only for those data points that will help it improve most efficiently.

This is especially useful when data labeling is expensive or labor-intensive. The active algorithm chooses examples with the highest uncertainty or those on the decision boundary.

Example: in a classification task, the model may select for labeling those objects about which it is least confident, thereby increasing the quality more quickly.

Active learning is closely related to weak supervision, where weak or incomplete sources of labels are used to train models.