FedDRLPD: Deep reinforcement Learning-Based defense mechanism against poisoning attacks in federated learning

Key Points

• Evidence suggests the implementation of deep reinforcement learning significantly reduces poisoning attacks.
• The analysis reported a reduction of over 30% in successful attacks when using the proposed mechanism.
• Assessment using simulations demonstrated the effectiveness of the federated learning framework and the deep reinforcement learning model.
• The findings highlight the need for advanced security protocols in federated learning to counteract ongoing threats.