Sine: Similarity is not enough for mitigating Local Model Poisoning Attacks in Federated Learning | IEEE Journals & Magazine | IEEE Xplore