9.3. Representations

Representations transform model outputs or feature spaces into vectors that can be matched against class-conditional summaries. They are used by compose and matching quantifiers to build the linear system or likelihood objective.

9.3.1. Core representations

• BaseRepresentation

• HistogramRepresentation

• KDERepresentation

• DistanceRepresentation

• KernelMeanRepresentation

• PredictionRepresentation

• HardPredictionRepresentation

• SoftPredictionRepresentation

9.3.2. Example

from mlquantify.representations import HistogramRepresentation

rep = HistogramRepresentation(bins=(10,))
rep.fit(train_scores, y_train)
test_representation = rep.transform(test_scores)