國立虎尾科技大學 |

Emulated Autoencoder: A Time-Efficient Image Denoiser for Defense of Convolutional Neural Networks Against Evasion Attacks.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Emulated Autoencoder: A Time-Efficient Image Denoiser for Defense of Convolutional Neural Networks Against Evasion Attacks./
作者:	Le, Dat Tien.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2022,
面頁冊數:	114 p.
附註:	Source: Masters Abstracts International, Volume: 84-01.
Contained By:	Masters Abstracts International84-01.
標題:	Computer science. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29211607
ISBN:	9798837529085

Emulated Autoencoder: A Time-Efficient Image Denoiser for Defense of Convolutional Neural Networks Against Evasion Attacks.
Le, Dat Tien.

Emulated Autoencoder: A Time-Efficient Image Denoiser for Defense of Convolutional Neural Networks Against Evasion Attacks. - Ann Arbor : ProQuest Dissertations & Theses, 2022 - 114 p.

Source: Masters Abstracts International, Volume: 84-01.

Thesis (Master's)--University of Washington, 2022.

This item must not be sold to any third party vendors.

As Convolutional Neural Networks (CNN) have become essential to modern applications such as image classification on social networks or self-driving vehicles, evasion attacks targeting CNNs can lead to damage for users. Therefore, there has been a rising amount of research focusing on defending against evasion attacks. Image denoisers have been used to mitigate the impact of evasion attacks; however, there is not a sufficiently broad view of the use of image denoisers as adversarial defenses in image classification due to a lack of trade-off analysis. Thus, image denoisers' costs, including training time, image reconstruction time, and loss of benign F1 scores of CNN classifiers, are explored in this thesis. Additionally, Emulated Autoencoder (EAE), which is the proposed method of this thesis to optimize trade-offs for high volume classification tasks, is evaluated alongside state-of-the-art image denoisers in the gray-box and white-box threat models. EAE outperforms most image denoisers in both the gray-box and white-box threat models while drastically reducing training and image reconstruction time compared to the state-of-the-art denoisers. As a result, EAE is more appropriate for securing high-volume classification applications of images.

ISBN: 9798837529085Subjects--Topical Terms:

573171
Computer science.
Subjects--Index Terms:

Adversarial machine learning

Emulated Autoencoder: A Time-Efficient Image Denoiser for Defense of Convolutional Neural Networks Against Evasion Attacks.
LDR:02507nam a2200373 4500 001 1104622
005 20230619080104.5
006 m o d
007 cr#unu||||||||
008 230907s2022 ||||||||||||||||| ||eng d
020 $a 9798837529085
035 $a (MiAaPQ)AAI29211607
035 $a AAI29211607
040 $a MiAaPQ $c MiAaPQ
100 1 $a Le, Dat Tien. $3 1413515
245 1 0 $a Emulated Autoencoder: A Time-Efficient Image Denoiser for Defense of Convolutional Neural Networks Against Evasion Attacks.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2022
300 $a 114 p.
500 $a Source: Masters Abstracts International, Volume: 84-01.
500 $a Advisor: Lagesse, Brent.
502 $a Thesis (Master's)--University of Washington, 2022.
506 $a This item must not be sold to any third party vendors.
520 $a As Convolutional Neural Networks (CNN) have become essential to modern applications such as image classification on social networks or self-driving vehicles, evasion attacks targeting CNNs can lead to damage for users. Therefore, there has been a rising amount of research focusing on defending against evasion attacks. Image denoisers have been used to mitigate the impact of evasion attacks; however, there is not a sufficiently broad view of the use of image denoisers as adversarial defenses in image classification due to a lack of trade-off analysis. Thus, image denoisers' costs, including training time, image reconstruction time, and loss of benign F1 scores of CNN classifiers, are explored in this thesis. Additionally, Emulated Autoencoder (EAE), which is the proposed method of this thesis to optimize trade-offs for high volume classification tasks, is evaluated alongside state-of-the-art image denoisers in the gray-box and white-box threat models. EAE outperforms most image denoisers in both the gray-box and white-box threat models while drastically reducing training and image reconstruction time compared to the state-of-the-art denoisers. As a result, EAE is more appropriate for securing high-volume classification applications of images.
590 $a School code: 0250.
650 4 $a Computer science. $3 573171
653 $a Adversarial machine learning
653 $a Autoencoder
653 $a Convolutional neural networks
653 $a Image denoiser
653 $a Machine learning adversarial defense
690 $a 0984
710 2 $a University of Washington. $b Computing and Software Systems. $3 1413516
773 0 $t Masters Abstracts International $g 84-01.
790 $a 0250
791 $a Master's
792 $a 2022
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29211607