… Adversary objective: In our attack settings, we perform targeted collusion attacks. We … strategy
to train the poisoning data locally and collude with other poisoning models. Meanwhile, the …

… To enhance the effectiveness and stealthiness of CMPA, we present the collusion-based
attack enhancement strategies, in which multiple malicious participants collaborate to construct …

… Recently, federated learning (FL) has emerged as an … 2) privacy attacks and defenses; and
3) poisoning attacks and defenses… For the collusion of Sybil attacks, contribution similarity [37] …

… proposed the Distributed Backdoor Attack (DBA), a novel colluding attack developed by fully
… 8) Federated Learning Environment Scope By conducting model training at the network …

… Federated Learning with partial participation, ie Np out of N participants are selected in each
training iteration of federated learning… to real-world machine learning applications, such as …

… This work focuses on defending against targeted model poisoning attacks in federated
learning. Without loss of generality, RoseAgg assumes that the data distributed across each …

… to resist collusion attacks. Our contributions are as follows. 1) We propose a federated learning
… modes and provides practical application scenarios for federated learning. 2) We use the …

… To protect clients’ data privacy against collusion attacks, we … server is involved in collusion
attacks. Moreover, considering that … to eliminate the effect of federation dynamics on FL. To the …

… We prove that PEFL provides postquantum security and guarantees aggregation oblivious
security considering collusion attacks. Performance evaluation indicates that PEFL has …

… Then we design a simple but effective score-based detection method to mitigate colluding
attacks. Third, we construct an enhanced spectral-based outlier detector to accurately discard …