Bandit Interpretability of Deep Models via Confidence Selection

Xiaoyue Duan, Hong Li, Panpan Wang, Tiancheng Wang, Boyu Liu, Baochang Zhang

Abstract: Highlights•We are the first to formulate the interpretation of deep models as a bandit problem.•Our statistical perturbations retain regional interaction without any prior involved.•The Upper Confidence Bound guarantees a fair selection of critical image features.•Our method provides more precise explanations with a smaller area.