Practical Adversarial Attacks Imperceptible to Humans in Visual Recognition  

作　　者：Donghyeok Park Sumin Yeon Hyeon Seo Seok-Jun Buu Suwon Lee 

机构地区：[1]Aircraft Final Assembly Manufacturing Engineering Team,Korea Aerospace Industries,Sacheon-si,52529,Republic of Korea [2]Department of Computer Science and Engineering,Gyeongsang National University,Jinju-si,52828,Republic of Korea

出　　处：《Computer Modeling in Engineering & Sciences》2025年第3期2725-2737,共13页工程与科学中的计算机建模(英文)

基　　金：supported by the Research Resurgence under the Glocal University 30 Project at Gyeongsang National University in 2024.

摘　　要：Recent research on adversarial attacks has primarily focused on white-box attack techniques,with limited exploration of black-box attack methods.Furthermore,in many black-box research scenarios,it is assumed that the output label and probability distribution can be observed without imposing any constraints on the number of attack attempts.Unfortunately,this disregard for the real-world practicality of attacks,particularly their potential for human detectability,has left a gap in the research landscape.Considering these limitations,our study focuses on using a similar color attack method,assuming access only to the output label,limiting the number of attack attempts to 100,and subjecting the attacks to human perceptibility testing.Through this approach,we demonstrated the effectiveness of black box attack techniques in deceiving models and achieved a success rate of 82.68%in deceiving humans.This study emphasizes the significance of research that addresses the challenge of deceiving both humans and models,highlighting the importance of real-world applicability.

关 键 词：Adversarial attacks image recognition information security 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]