Towards Unified, Practical Evaluation of Model Poisoning Attacks and Defence in Federated Learning | IEEE Conference Publication | IEEE Xplore