A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Filters
Poster: Fooling XAI with Explanation-Aware Backdoors
2023
Proceedings of the 2023 ACM SIGSAC Conference on Computer and Communications Security
Preserved Fulltext
While we have presented successful explanation-aware backdoors in our original work, "Disguising Attacks with Explanation-Aware Backdoors, " in this paper, we provide a brief overview and a focus on the ...
Explanation-aware backdoors therefore can bypass explanation-based detection techniques and "throw a red herring" at the human analyst. ...
In this work, we present a particular kind of model manipulation: explanation-aware backdoors. ...
doi:10.1145/3576915.3624379
fatcat:wl2cakt2yjhgfnv3vrbps5ak34
Backdooring Explainable Machine Learning
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
In this paper, we demonstrate blinding attacks that can fully disguise an ongoing attack against the machine learning model. ...
Similar to neural backdoors, we modify the model's prediction upon trigger presence but simultaneously also fool the provided explanation. ...
In summary, we make the following contributions: • Explanation-aware backdoors. ...
arXiv:2204.09498v1
fatcat:7ebxccwfwfcqviypeewrqwoyx4
Revealing Vulnerabilities of Neural Networks in Parameter Learning and Defense Against Explanation-Aware Backdoors
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
The method we suggest defences against most modern explanation-aware adversarial attacks, achieving an approximate decrease of ~99\% in the Attack Success Rate (ASR) and a ~91\% reduction in the Mean Square ...
Error (MSE) between the original explanation and the defended (post-attack) explanation across three unique types of attacks. ...
Revealing Vulnerabilities of Neural Networks in Parameter Learning and Defense Against Explanation-Aware Backdoors Supplementary Material
Attack Scenarios Table 6 presents various attack and evaluation ...
arXiv:2403.16569v1
fatcat:bbc6ecnk5vevjnpf55oy76ozja
WaNet – Imperceptible Warping-based Backdoor Attack
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
With the thriving of deep learning and the widespread practice of using pre-trained networks, backdoor attacks have become an increasing security threat drawing many research interests in recent years. ...
Behavior analyses show that our backdoors are transparent to network inspection, further proving this novel attack mechanism's efficiency. ...
Lately, Liu et al. (2020) proposed to disguise backdoor triggers as reflectance to make the poisoned images look natural. ...
arXiv:2102.10369v4
fatcat:ye5s7eye55a4ldn4dtf3bozpyy
Backdoor Attacks and Countermeasures on Deep Learning: A Comprehensive Review
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Other Versions
This work provides the community with a timely comprehensive review of backdoor attacks and countermeasures on deep learning. ...
In some cases, an attacker can intelligently bypass existing defenses with an adaptive attack. ...
New devised empirical countermeasures against backdoor attacks would soon be broken by strong adaptive backdoor attacks when the attacker is aware of the defense. ...
arXiv:2007.10760v3
fatcat:6i4za6345jedbe5ek57l2tgld4
ASSET: Robust Backdoor Data Detection Across a Multiplicity of Deep Learning Paradigms
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Other Versions
Successful backdoor attacks have also been demonstrated in these new settings. ...
attack. ...
Figure 7 : 7 Figure 7: Visual examples of the backdoor poisoned samples disguised by adaptive attacks (Case-0, CIFAR-10). ...
arXiv:2302.11408v2
fatcat:5hg666to2naz5ipe35hsfxkoku
Backdoor Attack with Mode Mixture Latent Modification
[article]
2024
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2024; you can also visit the original URL.
The file type is application/pdf.
aware of the existence of backdoor attacks. ...
Previous research can be categorized into two genres: poisoning a portion of the dataset with triggered images for users to train the model from scratch, or training a backdoored model alongside a triggered ...
A potential explanation is that previous methods for backdoor attacks depend on θe to learn to cluster poisoned images in the latent space and then classify them using θc. ...
arXiv:2403.07463v1
fatcat:iwgzfqaxyvc3tovpbeg3ycvqae
Poison Attack and Defense on Deep Source Code Processing Models
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
The attackers aim to inject insidious backdoors into models by poisoning the training data with poison samples. ...
By activating backdoors, attackers can manipulate the poisoned models in security-related scenarios. ...
Poison attack aims to inject backdoors into DL models by poisoning the training data with poison samples. ...
arXiv:2210.17029v1
fatcat:bi3to7ta5bbtlc5iqo2qzwwjcm
Review on Malware, Types, and its Analysis
2022
International Journal for Research in Applied Science and Engineering Technology
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Malicious software, sometimes known as malware or mal-ware, is a type of program with malicious intents. ...
Malware that compromises computers through a coordinated attack includes worms, trojan horses, botnets, and rootkits. ...
The purpose of the disguised payloads or attacks is to deceive the user into carrying out execution by making them appear friendly. ...
doi:10.22214/ijraset.2022.47887
fatcat:qfehxarjkzhlzdfccrzwyibqam
Machine Learning Security: Threats, Countermeasures, and Evaluations
2020
IEEE Access
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
INDEX TERMS Artificial intelligence security, poisoning attacks, backdoor attacks, adversarial examples, privacy-preserving machine learning. ...
It has demonstrated significant success in dealing with various complex problems, and shows capabilities close to humans or even beyond humans. ...
The malicious workers can be disguised as normal ones to evade detection, while achieving the maximum attack utility meanwhile. ...
doi:10.1109/access.2020.2987435
fatcat:ksinvcvcdvavxkzyn7fmsa27ji
Page 194 of The Academy and Literature Vol. 7, Issue 146
[page]
1875
The Academy and Literature
Preserved Fulltext
The Internet Archive has digitized a microfilm copy of this work. It may be possible to borrow a copy for reading.
The name “ Moro” might very naturally suggest to Cinthio the desire of disguising his hero as a Moor—a device which would save the author from unpleasant consequences, while the disguise would be too flimsy ...
(Vol. ii. pp. 141, 142, 166,
That in his defence Rowland Williams adopted a “policy of evasion,” and availed himself of “ —— and backdoors,” is simply untrue. ...
Fully Undetectable Remote Access Trojan: Android
2019
International Journal for Research in Applied Science and Engineering Technology
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
acting as the interface between the server and the attacker. ...
With over 80 percent market share, Android is the most prevailing player in the mobile platform and people are switching to it with its every update. ...
A computer with a sophisticated backdoor Trojan installed may also be referred to as a zombie or bot. ...
doi:10.22214/ijraset.2019.5182
fatcat:yuh25mmekne37d5bab5gxblc5u
Towards Security Threats of Deep Learning Systems: A Survey
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Other Versions
In particular, we focus on four types of attacks associated with security threats of deep learning: model extraction attack, model inversion attack, poisoning attack and adversarial attack. ...
To this end, lots of research has been conducted with the purpose of exhaustively identifying intrinsic weaknesses and subsequently proposing feasible mitigation. ...
In recent years, with the development of technology, more research has focused on backdoor poisoning attacks [258] [76] [141] [23] [212] . ...
arXiv:1911.12562v2
fatcat:m3lyece44jgdbp6rlcpj6dz2gm
Encryption and Globalization
2011
Social Science Research Network
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2011; you can also visit the original URL.
The file type is application/pdf.
The apparent source of attack is often not the actual source. 86 The ability to disguise the source of an attack greatly weakens deterrence, because the defense often has no workable way to locate and ...
; 2) attacks that are more efficient than brute force; and 3) attacks assisted by a flaw known to the attacker, or "backdoors." ...
doi:10.2139/ssrn.1960602
fatcat:apqj5d2f4ra2jfi46ei4ruuqg4
Strategically-Motivated Advanced Persistent Threat: Definition, Process, Tactics and a Disinformation Model of Counterattack
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Finally, we present a general disinformation model, derived from situation awareness theory, and explain how disinformation can be used to attack the situation awareness and decision making of not only ...
APT refers to knowledgeable human attackers that are organized, highly sophisticated and motivated to achieve their objectives against a targeted organization(s) over a prolonged period. ...
Some intrusions have been facilitated through the distribution of hardware with built-in backdoors-"malicious firmware" (Sood and Enbody 2013). ...
arXiv:2103.15005v1
fatcat:ihrpyqzsinemrauntrf2mquqgm
« Previous
Showing results 1 — 15 out of 495 results