Hidden Trigger Backdoor Attacks
نویسندگان
چکیده
منابع مشابه
Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning
Deep learning models have achieved high performance on many tasks, and thus have been applied to many security-critical scenarios. For example, deep learning-based face recognition systems have been used to authenticate users to access many security-sensitive applications like payment apps. Such usages of deep learning systems provide the adversaries with sufficient incentives to perform attack...
متن کاملBackdoor Trees
The surprisingly good performance of modern satisfiability (SAT) solvers is usually explained by the existence of a certain “hidden structure” in real-world instances. We introduce the notion of backdoor trees as an indicator for the presence of a hidden structure. Backdoor trees refine the notion of strong backdoor sets, taking into account the relationship between backdoor variables. We prese...
متن کاملBackdoor Branching
We present an exact Mixed Integer Programming (MIP) solution scheme where a set covering model is used to find a small set of first-choice branching variables. In a preliminary “sampling” phase, our method quickly collects a number of relevant lowcost fractional solutions that qualify as obstacles for the Linear Programming (LP) relaxation bound improvement. Then a set covering model is solved ...
متن کاملFrom Horn Strong Backdoor Sets to Ordered Strong Backdoor Sets
Identifying and exploiting hidden problem structures is recognized as a fundamental way to deal with the intractability of combinatorial problems. Recently, a particular structure called (strong) backdoor has been identified in the context of the satisfiability problem. Connections has been established between backdoors and problem hardness leading to a better approximation of the worst case ti...
متن کاملCountering Hidden-Action Attacks on Networked Systems
We define an economic category of hidden-action attacks: actions made attractive by a lack of observation. We then consider its implications for computer systems. Rather than structure contracts to compensate for incentive problems, we rely on insights from social capital theory to design network topologies and interactions that undermine the potential for hidden-action attacks.
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the AAAI Conference on Artificial Intelligence
سال: 2020
ISSN: 2374-3468,2159-5399
DOI: 10.1609/aaai.v34i07.6871