ActiveLearningCandidate<T>

Models & Types · AiDotNet.ActiveLearning

One pool sample’s place in an active-learning ranking.

For Beginners

Active learning ranks unlabeled samples by how useful it would be to label them next. This is one sample’s entry in that ranking.

Constructors

Constructor Summary
ActiveLearningCandidate()

Properties

Property Summary
T Informativeness { get; set; } Gets the strategy’s raw informativeness (uncertainty) score for this sample.
T MarginalGain { get; set; } Gets the diversity-adjusted score at the moment this sample was chosen (informativeness minus its redundancy against the already-chosen batch), or the raw informativeness for unselected samples.
int PoolIndex { get; set; } Gets the sample’s index in the unlabeled pool.
int SelectionOrder { get; set; } Gets the order this sample was chosen into the batch (0 = chosen first), or -1 if it was not selected into the batch.