calibration layer

Go back to the [[AI Glossary]]

A post-prediction adjustment, typically to account for prediction bias. The adjusted predictions and probabilities should match the distribution of an observed set of labels.