Explainability Methods for Deep Neural Networks

Comparative study of LIME, SHAP, Grad-CAM, and integrated gradients across three classification tasks.