Machine learning interpretability methods

Post-hoc and model-agnostic techniques that explain the predictions of machine learning models by attributing importance to input features, including SHAP values, saliency maps, and LIME.

AI-generated definition