Reference : Adversarial Robustness in Multi-Task Learning: Promises and Illusions

	Document type :	Scientific congresses, symposiums and conference proceedings : Paper published in a book
	Discipline(s) :	Engineering, computing & technology : Computer science
	Focus Areas :	Security, Reliability and Trust
	To cite this reference:	http://hdl.handle.net/10993/48724

Title :	Adversarial Robustness in Multi-Task Learning: Promises and Illusions
Language :	English
Author, co-author :	Ghamizi, Salah [University of Luxembourg > Faculty of Science, Technology and Medicine (FSTM) > Department of Computer Science (DCS) >]
	Cordy, Maxime [University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > SerVal >]
	Papadakis, Mike [University of Luxembourg > Faculty of Science, Technology and Medicine (FSTM) > Department of Computer Science (DCS) >]
	Le Traon, Yves [University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT) > SerVal >]
Publication date :	2022
Main document title :	Proceedings of the thirty-Sixth AAAI Conference on Artificial Intelligence (AAAI-22)
Peer reviewed :	Yes
On invitation :	No
Audience :	International
ISBN :	1-57735-876-7
Event name :	The Thirty-Sixth AAAI Conference on Artificial Intelligence (AAAI-22)
Event date :	from 22-02-2022 to 01-03-2022
Event organizer :	AAAI
Event place (city) :	Virtual
Keywords :	[en] Robustness ; Deep Learning ; Multi-task
Abstract :	[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.
Target :	Researchers ; Professionals
Permalink :	http://hdl.handle.net/10993/48724
DOI :	10.1609/aaai.v36i1.19950
FnR project :	FnR ; FNR12669767 > Yves Le Traon > STELLAR > Testing Self-learning Systems > 01/09/2019 > 31/08/2022 > 2018

	File(s) associated to this reference
	Fulltext file(s): File Commentary Version Size Access Open access AAAI22_MT_aknowledgment.pdf Publisher postprint 305.25 kB View/Open Additional material(s): File Commentary Size Access Open access AAAI22_MT_Supp.pdf 2.65 MB View/Open

All documents in ORBi^lu are protected by a user license.

 

University of Luxembourg Library |  Feedback |  Legal notices Site Map