How the softmax output is misleading for evaluating the strength of adversarial examples

Poster (Scientific congresses and symposiums)

Ozbulak, Utku; De Neve, Wesley; Van Messem, Arnout

2018 • 32nd Conference on Neural Information Processing Systems (NeurIPS): Workshop on Security in Machine Learning (SECML)

Permalink
https://hdl.handle.net/2268/265534

Utku_Ozbulak_SECML18_Poster.pdf

Publisher postprint (664.33 kB)

All documents in ORBi are protected by a user license.

Disciplines :

Computer science

Author, co-author :

Ozbulak, Utku

De Neve, Wesley

Van Messem, Arnout ; Ghent University Global Campus

Language :

English

Title :

How the softmax output is misleading for evaluating the strength of adversarial examples

Publication date :

December 2018

Event name :

32nd Conference on Neural Information Processing Systems (NeurIPS): Workshop on Security in Machine Learning (SECML)

Event place :

Montréal, Canada

Event date :

Dec 3-8, 2018

Audience :

International

Available on ORBi :

since 29 November 2021

Number of views

43 (1 by ULiège)

Number of downloads

12 (0 by ULiège)