Why ReLU networks yield high-confidence predictions far away from the training data and how to mitigate the problem

DSpace Repositorium (Manakin basiert)

Why ReLU networks yield high-confidence predictions far away from the training data and how to mitigate the problem

Autor(en): Hein, Matthias; Andriushchenko, Maksym; Bitterwolf, Julian
Tübinger Autor(en):
Hein, Matthias
Bitterwolf, Julian
Erschienen in: 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2019), Bd. , S. 41-50
Sprache: Englisch
Referenz zum Volltext: http://dx.doi.org/10.1109/CVPR.2019.00013
ISBN: 978-1-7281-3293-8
DDC-Klassifikation: 004 - Informatik
Dokumentart: Wissenschaftlicher Artikel
Konferenzveröffentlichung
Zur Langanzeige

Das Dokument erscheint in: