Explainable AI for image classification models: saliency maps and interpretable classification

par Ronan Sicre (IRIT)

mardi 25 nov. 2025, 11:15 → 12:15 Europe/Paris

Salle K. Johnson (1R3, 1er étage)

We will first review previous work on explainable AI methods and present two works on this topic. We will first present Opti-CAM, a CAM-based method that optimizes a masking objective per instance. The resulting saliency map highlights the important area of an image regarding the decision of a trained image classification network. Then, we will focus on image classification using parts or prototype-based architectures. Such model is explainable by design and brings transparency to a model.