Demonstrates malicious model updates and backdoor behavior under federated-learning attack assumptions.
Sources retain their original publishers’ rights. Inclusion does not imply endorsement. Saved sources stay in this browser on this device. Dates and source descriptions come from the existing editorial records.