[20p-A308-1~20] CS.2 Code-sharing Session of 3.2 & 4.4

2:45 PM - 3:00 PM JST(5:45 AM - 6:00 AM UTC)

[20p-A308-8] Effects of Trigger Size on Data Poisoning Attack in Imaging through Scattering Media by Deep Learning

〇Daiki Shingu¹, Wataru Watanabe¹ (1.Ritsumeikan Univ.)

Keywords:

deep learning,vulnerability

Scattering media such as fog and frosted glass diffuse transmitted light, making it difficult to recognize objects behind scattering media. Recently, deep learning has been attracting attention as a method for reconstructing objects behind scattering media. However, deep learning is vulnerable to adversarial examples and data poisoning attacks. In this study, we investigate the effect of trigger size on data poisoning attacks in imaging through scattering media.