Clean-label poisoning attack with perturbation causing dominant features

Open Webpage

Chen Zhang, Zhuo Tang, Kenli Li

Published: 2023, Last Modified: 07 Jan 2026Inf. Sci. 2023EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•A clean-label poisoning attack against DNNs.•Expand our data poisoning attack into a powerful backdoor attack.•Imperceptible but dominant image features.