Improving Prototypical Part Networks with Reward Reweighing, Reselection, and Retraining

Open in new window