東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Enhancing Deep Learning Security Thr...

Kiourti, Panagiota,

Linked to FindBook

Google Book

Amazon

博客來

Enhancing Deep Learning Security Through Explainability and Robustness /

Record Type:	Electronic resources : Monograph/item
Title/Author:	Enhancing Deep Learning Security Through Explainability and Robustness // Panagiota Kiourti.
Author:	Kiourti, Panagiota,
Description:	1 electronic resource (130 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 86-03, Section: B.
Contained By:	Dissertations Abstracts International86-03B.
Subject:	Computer engineering. -
Online resource:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31487737
ISBN:	9798384076735

Enhancing Deep Learning Security Through Explainability and Robustness /
Kiourti, Panagiota,

Enhancing Deep Learning Security Through Explainability and Robustness /Panagiota Kiourti. - 1 electronic resource (130 pages)

Source: Dissertations Abstracts International, Volume: 86-03, Section: B.

The growing interest in deploying deep learning models in critical applications has raised concerns about their vulnerabilities, particularly to backdoor or Trojan attacks. These attacks aim to train a network to respond maliciously to specially crafted trigger patterns in the inputs while exhibiting state-of-the-art performance. This thesis addresses the identification of such attacks in deep reinforcement learning, proposes a mitigation strategy for their detection in classification neural networks in production using feature attribution methods, and introduces a new framework for evaluating the robustness of attribution methods.Firstly, TrojDRL is introduced as a tool for exploring and evaluating backdoor attacks on deep reinforcement learning agents. TrojDRL exploits the sequential nature of deep reinforcement learning (DRL) and considers various threat model gradations. It introduces untargeted attacks on state-of-the-art actor-critic policy networks that can circumvent existing defenses built on the assumption that backdoors are targeted. TrojDRL shows that the attacks require only as little as 0.025% poisoning of the training data. Compared with existing works of backdoor attacks on classification models, this tool is a pioneering effort toward understanding the vulnerability of DRL agents.Secondly, this thesis presents MISA, a new online detection approach for Trojan triggers present in neural networks at inference time after the deployment of the model. MISA utilizes feature attribution methods to explain the decision of a neural network. It defines misattributions to capture the anomalous manifestation of a Trojan activation in the feature attribution space by first computing the input's attribution on different features and then statistically analyzing these attributions to ascertain the presence of a Trojan trigger. Across a set of benchmarks, MISA can effectively detect Trojan triggers for a wide variety of trigger patterns, achieving 96\\% AUC for detecting Trojan-triggered images without any assumptions on the trigger pattern.Lastly, the robustness of feature attribution methods for deep neural networks is critically examined. This thesis challenges the current notion of attributional robustness that largely ignores the difference in the model's outputs and introduces a new evaluation framework. This involves defining similar inputs in a different way than existing methods do and introducing a novel method based on generative adversarial networks to generate these inputs, leading to a different definition of attributional robustness. The new robustness metric is comprehensively evaluated against existing metrics and state-of-the-art attribution methods. The findings highlight the need for a more objective metric that reveals the weaknesses of an attribution method rather than that of the neural network, thus providing a more accurate evaluation of the robustness of attribution methods.

English

ISBN: 9798384076735Subjects--Topical Terms:

621879
Computer engineering.
Subjects--Index Terms:

Deep learning

Enhancing Deep Learning Security Through Explainability and Robustness /
LDR:04418nmm a22004453i 4500 001 2429376
005 20250430100812.5
006 m o d
007 cr|nu||||||||
008 261011s2024 miu||||||m |||||||eng d
020 $a 9798384076735
035 $a (MiAaPQD)AAI31487737
035 $a AAI31487737
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Kiourti, Panagiota, $e author. $0 (orcid)0009-0006-8538-901X $3 3817193
245 1 0 $a Enhancing Deep Learning Security Through Explainability and Robustness / $c Panagiota Kiourti.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2024
300 $a 1 electronic resource (130 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 86-03, Section: B.
500 $a Advisors: Li, Wenchao Committee members: Stringhini, Gianluca; Ohn-bar, Eshed; Batmanghelich, Kayhan.
502 $b Ph.D. $c Boston University $d 2024.
520 $a The growing interest in deploying deep learning models in critical applications has raised concerns about their vulnerabilities, particularly to backdoor or Trojan attacks. These attacks aim to train a network to respond maliciously to specially crafted trigger patterns in the inputs while exhibiting state-of-the-art performance. This thesis addresses the identification of such attacks in deep reinforcement learning, proposes a mitigation strategy for their detection in classification neural networks in production using feature attribution methods, and introduces a new framework for evaluating the robustness of attribution methods.Firstly, TrojDRL is introduced as a tool for exploring and evaluating backdoor attacks on deep reinforcement learning agents. TrojDRL exploits the sequential nature of deep reinforcement learning (DRL) and considers various threat model gradations. It introduces untargeted attacks on state-of-the-art actor-critic policy networks that can circumvent existing defenses built on the assumption that backdoors are targeted. TrojDRL shows that the attacks require only as little as 0.025% poisoning of the training data. Compared with existing works of backdoor attacks on classification models, this tool is a pioneering effort toward understanding the vulnerability of DRL agents.Secondly, this thesis presents MISA, a new online detection approach for Trojan triggers present in neural networks at inference time after the deployment of the model. MISA utilizes feature attribution methods to explain the decision of a neural network. It defines misattributions to capture the anomalous manifestation of a Trojan activation in the feature attribution space by first computing the input's attribution on different features and then statistically analyzing these attributions to ascertain the presence of a Trojan trigger. Across a set of benchmarks, MISA can effectively detect Trojan triggers for a wide variety of trigger patterns, achieving 96\\% AUC for detecting Trojan-triggered images without any assumptions on the trigger pattern.Lastly, the robustness of feature attribution methods for deep neural networks is critically examined. This thesis challenges the current notion of attributional robustness that largely ignores the difference in the model's outputs and introduces a new evaluation framework. This involves defining similar inputs in a different way than existing methods do and introducing a novel method based on generative adversarial networks to generate these inputs, leading to a different definition of attributional robustness. The new robustness metric is comprehensively evaluated against existing metrics and state-of-the-art attribution methods. The findings highlight the need for a more objective metric that reveals the weaknesses of an attribution method rather than that of the neural network, thus providing a more accurate evaluation of the robustness of attribution methods.
546 $a English
590 $a School code: 0017
650 4 $a Computer engineering. $3 621879
650 4 $a Computer science. $3 523869
650 4 $a Systems science. $3 3168411
653 $a Deep learning
653 $a Deep reinforcement learning
653 $a Trojan triggers
653 $a Robustness
653 $a Attribution methods
690 $a 0464
690 $a 0800
690 $a 0984
690 $a 0790
710 2 $a Boston University. $b Electrical & Computer Engineering ENG. $e degree granting institution. $3 3770549
720 1 $a Li, Wenchao $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 86-03B.
790 $a 0017
791 $a Ph.D.
792 $a 2024
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31487737