Tag
The paper proposes FRAC-MARL, a decentralized actor-critic multi-agent reinforcement learning method that achieves full Byzantine resilience by leveraging redundancy in communication, ensuring convergence to optimal parameters even under adversarial attacks.
This paper proposes a quantum annealing approach that reformulates client selection in federated learning as a QUBO problem to defend against Byzantine attacks, showing improved detection accuracy over classical MultiKrum on sophisticated attacks, especially when combined with a MultiSignal ensemble.