Reference : Adversarial Robustness in Multi-Task Learning: Promises and Illusions
E-prints/Working papers : First made available on ORBilu
Engineering, computing & technology : Computer science
Security, Reliability and Trust
http://hdl.handle.net/10993/48724
Adversarial Robustness in Multi-Task Learning: Promises and Illusions
English
Ghamizi, Salah mailto [University of Luxembourg > Faculty of Science, Technology and Medicine (FSTM) > Department of Computer Science (DCS) >]
Cordy, Maxime mailto [University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > SerVal >]
Papadakis, Mike mailto [University of Luxembourg > Faculty of Science, Technology and Medicine (FSTM) > Department of Computer Science (DCS) >]
Le Traon, Yves mailto [University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > SerVal >]
2021
No
[en] Robustness ; Deep Learning ; Multi-task
[en] Vulnerability to adversarial attacks is a well-known weakness of Deep Neural networks. While most of the studies focus on single-task neural networks with computer vision datasets, very little research has considered complex multi-task models that are common in real applications. In this paper, we evaluate the design choices that impact the robustness of multi-task deep learning networks. We provide evidence that blindly adding auxiliary tasks, or weighing the tasks provides a false sense of robustness. Thereby, we tone down the claim made by previous research and study the different factors which may affect robustness. In particular, we show that the choice of the task to incorporate in the loss function are important factors that can be leveraged to yield more robust models.
http://hdl.handle.net/10993/48724

File(s) associated to this reference

Fulltext file(s):

FileCommentaryVersionSizeAccess
Open access
AAAI22_MT.pdfAuthor preprint302.75 kBView/Open

Additional material(s):

File Commentary Size Access
Open access
AAAI22_MT_Supp.pdf2.65 MBView/Open

Bookmark and Share SFX Query

All documents in ORBilu are protected by a user license.