ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness

DSpace Repository

ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness

Author: Geirhos, Robert; Rubisch, Patricia; Michaelis, Claudio; Bethge, Matthias; Wichmann, Felix A.; Brendel, Wieland
Tübinger Autor(en):
Bethge, Matthias
Wichmann, Felix A.
Geirhos, Robert
Rubisch, Patricia
Michaelis, Claudio
Brendel, Wieland
Published in: arXiv (2018-11-29), Bd. 1811.12231
Language: English
Full text: https://arxiv.org/abs/1811.12231
DDC Classifikation: 004 - Data processing and computer science
Dokumentart: Preprint
Show full item record

This item appears in the following Collection(s)