Hypernym Bias: Unraveling Deep Classifier Training Dynamics through the Lens of Class Hierarchy.

Roman Malashin, Valeria Yachnaya, Alexandr V. Mullin

Browse the full AISTATS paper archive.