QueryStrategy¶
- class probly.evaluation.active_learning.strategies.QueryStrategy(*args, **kwargs)[source]¶
-
Protocol for active learning query strategies.
- select(estimator: E, pool: ActiveLearningPool, n: int) ArrayLike[source]¶
Return n indices into pool.x_unlabeled to query next.
- Parameters:
estimator – A fitted estimator used to score unlabeled samples.
pool – The current active learning pool.
n – Number of samples to select.
- Returns:
Array of n integer indices into pool.x_unlabeled.