UncertaintySamplingStrategy<T, TInput, TOutput>

Models & Types · AiDotNet.ActiveLearning.Strategies.Uncertainty

Uncertainty sampling strategy for active learning.

Implements: IUncertaintyStrategy<T, TInput, TOutput>

For Beginners

Uncertainty sampling is the simplest and most popular active learning strategy. It selects samples where the model is most uncertain about its prediction.

How It Works

Uncertainty Measures:

When to Use:

Reference: Lewis and Gale “A Sequential Algorithm for Training Text Classifiers” (1994)

Constructors

Constructor Summary
UncertaintySamplingStrategy() Initializes a new uncertainty sampling strategy with default entropy measure.
UncertaintySamplingStrategy(UncertaintyMeasure uncertaintyMeasure, ActiveLearnerConfig<T> config) Initializes a new uncertainty sampling strategy with specified measure.

Properties

Property Summary
string Description { get; }
string Name { get; }

Methods

Method Summary
T ComputeUncertainty(IFullModel<T, TInput, TOutput> model, TInput input)
Vector<T> ComputeScores(IFullModel<T, TInput, TOutput> model, IDataset<T, TInput, TOutput> unlabeledPool)
Vector<T> GetPredictionProbabilities(IFullModel<T, TInput, TOutput> model, TInput input)
int[] SelectSamples(IFullModel<T, TInput, TOutput> model, IDataset<T, TInput, TOutput> unlabeledPool, int batchSize)
void Reset()
void UpdateState(int[] newlyLabeledIndices, TOutput[] labels)