Clean-image Backdoor: Attacking Multi-label Models with Poisoned Labels Only

Multi-label learning Multi-label learning is commonly exploited to recognize a set of categories in an...