핵심 개념
Data poisoning attacks in federated learning can significantly impact server accuracy and ASR, with LF attacks failing and FP attacks proving effective.
통계
LF 공격은 서버 정확도가 0.0428이고 ASR이 0.9564임.
FP 공격은 서버 정확도와 ASR이 모두 약 0.9600임.
인용구
"LF attacks failed to fool the server, while FP attacks were successful in deceiving the system."