IQueryStrategy<T, TInput, TOutput>

Interfaces · AiDotNet.ActiveLearning.Interfaces

Interface for query strategies in active learning.

Known implementations: BADGEStrategy<T, TInput, TOutput>, IBayesianStrategy<T, TInput, TOutput>, ICommitteeStrategy<T, TInput, TOutput>, IDensityWeightedStrategy<T, TInput, TOutput>, IUncertaintyStrategy<T, TInput, TOutput>

For Beginners

A query strategy determines which unlabeled samples should be selected for labeling by the oracle (human expert). Different strategies use different criteria for selection.

How It Works

Common Query Strategies:

Properties

Property Summary
string Description { get; } Gets a description of how the strategy works.
string Name { get; } Gets the name of the query strategy.

Methods

Method Summary
Vector<T> ComputeScores(IFullModel<T, TInput, TOutput> model, IDataset<T, TInput, TOutput> unlabeledPool) Computes informativeness scores for all samples in the unlabeled pool.
int[] SelectSamples(IFullModel<T, TInput, TOutput> model, IDataset<T, TInput, TOutput> unlabeledPool, int batchSize) Selects the most informative samples from the unlabeled pool.
void Reset() Resets the strategy to its initial state.
void UpdateState(int[] newlyLabeledIndices, TOutput[] labels) Updates the strategy’s internal state after new samples are labeled.