1 paper · 1 filter
Tores Julie, Sun Rémy, Sassatelli Lucile +3
Concept-based (CB) models provide interpretability and support test-time human intervention, while standard neural networks (NN) offer strong task performance but little transparen…