1 paper
Utku Åirin, Cathy Hou, David Alvarez-Melis +1
Neural networks suffer from shortcut learning, where learned features generalize well to the training set but not to in-distribution (ID) or out-of-distribution (OOD) test sets. Ex…