Everything tagged #interpretability

Explainable AI: Opening the Black Box with SHAP and LIME

High test accuracy is meaningless if a model exploits spurious correlations. How cooperative game theory and local linear surrogates bring mathematical interpretability to deep learning.