ITE Technical Group Submission System
Conference Paper's Information
Online Proceedings
[Sign in]
 Go Top Page Go Previous   [Japanese] / [English] 

Paper Abstract and Keywords
Presentation 2025-06-05 15:50
Comparative Analysis of Attack and Defense Methods Against Learned Image Compression Models
Jun Kurihara, Heming Sun (YNU)
Abstract (in Japanese) (See Japanese page) 
(in English) In recent years, learned image compression (LIC) models have attracted attention as a promising approach for achieving high-efficiency image compression. However, a key challenge of such machine learning-based models is their vulnerability to adversarial perturbations, which can significantly degrade the quality of the reconstructed images. In this study, we evaluate the performance of an adversarial attack using the Fast Gradient Sign Method (FGSM) as well as a defense method based on adversarial training. For the attack, we define three different loss functions based on the combinations of the original, perturbed, and reconstructed images, and compare the attack performance of FGSM under each loss. The results show that using the loss between the original and reconstructed images for gradient computation yields the most effective attack. For the defense, adversarial training is applied to the LIC model using adversarial examples. This approach successfully mitigates the degradation of the reconstructed images and improves the rate-distortion performance by 11.4% compared to the model before training.
Keyword (in Japanese) (See Japanese page) 
(in English) LIC model / Adversarial Attack / Adversarial Training / FGSM / / / /  
Reference Info. ITE Tech. Rep.
Paper #  
Date of Issue  
ISSN Online edition: ISSN 2424-1970
Download PDF

Conference Information
Committee IST ME IEICE-IE IEICE-BioX IEICE-SIP IEICE-MI  
Conference Date 2025-06-05 - 2025-06-06 
Place (in Japanese) (See Japanese page) 
Place (in English)  
Topics (in Japanese) (See Japanese page) 
Topics (in English)  
Paper Information
Registration To IEICE-IE 
Conference Code 2025-06-IST-ME-IE-BioX-SIP-MI 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Comparative Analysis of Attack and Defense Methods Against Learned Image Compression Models 
Sub Title (in English)  
Keyword(1) LIC model  
Keyword(2) Adversarial Attack  
Keyword(3) Adversarial Training  
Keyword(4) FGSM  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Jun Kurihara  
1st Author's Affiliation Yokohama National University (YNU)
2nd Author's Name Heming Sun  
2nd Author's Affiliation Yokohama National University (YNU)
3rd Author's Name  
3rd Author's Affiliation ()
4th Author's Name  
4th Author's Affiliation ()
5th Author's Name  
5th Author's Affiliation ()
6th Author's Name  
6th Author's Affiliation ()
7th Author's Name  
7th Author's Affiliation ()
8th Author's Name  
8th Author's Affiliation ()
9th Author's Name  
9th Author's Affiliation ()
10th Author's Name  
10th Author's Affiliation ()
11th Author's Name  
11th Author's Affiliation ()
12th Author's Name  
12th Author's Affiliation ()
13th Author's Name  
13th Author's Affiliation ()
14th Author's Name  
14th Author's Affiliation ()
15th Author's Name  
15th Author's Affiliation ()
16th Author's Name  
16th Author's Affiliation ()
17th Author's Name  
17th Author's Affiliation ()
18th Author's Name  
18th Author's Affiliation ()
19th Author's Name  
19th Author's Affiliation ()
20th Author's Name  
20th Author's Affiliation ()
21st Author's Name  
21st Author's Affiliation ()
22nd Author's Name  
22nd Author's Affiliation ()
23rd Author's Name  
23rd Author's Affiliation ()
24th Author's Name  
24th Author's Affiliation ()
25th Author's Name  
25th Author's Affiliation ()
26th Author's Name / /
26th Author's Affiliation ()
()
27th Author's Name / /
27th Author's Affiliation ()
()
28th Author's Name / /
28th Author's Affiliation ()
()
29th Author's Name / /
29th Author's Affiliation ()
()
30th Author's Name / /
30th Author's Affiliation ()
()
31st Author's Name / /
31st Author's Affiliation ()
()
32nd Author's Name / /
32nd Author's Affiliation ()
()
33rd Author's Name / /
33rd Author's Affiliation ()
()
34th Author's Name / /
34th Author's Affiliation ()
()
35th Author's Name / /
35th Author's Affiliation ()
()
36th Author's Name / /
36th Author's Affiliation ()
()
Speaker Author-1 
Date Time 2025-06-05 15:50:00 
Presentation Time 25 minutes 
Registration for IEICE-IE 
Paper #  
Volume (vol) vol.49 
Number (no)  
Page  
#Pages  
Date of Issue  


[Return to Top Page]

[Return to ITE Web Page]


The Institute of Image Information and Television Engineers (ITE), Japan