A single malicious client can craft a persistent, stealthy backdoor trigger for federated learning by simulating divergent local training paths and optimizing the trigger for consensus across those paths.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
DOPA: Stealthy and Generalizable Backdoor Attacks from a Single Client under Challenging Federated Constraints
A single malicious client can craft a persistent, stealthy backdoor trigger for federated learning by simulating divergent local training paths and optimizing the trigger for consensus across those paths.