Adaptive Federated Learning Defences via Trust-Aware Deep Q-Networks

Federated learning is vulnerable to poisoning and backdoor attacks under partial observability. We formulate defence as a partially observable sequential decision problem and introduce a trust-aware Deep Q-Network that integrates multi-signal evidence into client trust updates while optimizing a long-horizon robustness--accuracy objective. On CIFAR-10, we (i) establish a baseline showing steadily improving accuracy, (ii) show through a Dirichlet sweep that increased client overlap consistently improves accuracy and reduces ASR with stable detection, and (iii) demonstrate in a signal-budget study that accuracy remains steady while ASR increases and ROC-AUC declines as observability is reduced, which highlights that sequential belief updates mitigate weaker signals. Finally, a comparison with random, linear-Q, and policy gradient controllers confirms that DQN achieves the best robustness--accuracy trade-off.

Key Contributions

• First POMDP formulation of FL defense, treating client trust as latent state and aggregation as sequential decision-making under partial observability
• Multi-signal Bayesian trust tracking pipeline combining directional alignment, magnitude deviation, and validation impact across rounds
• Trust-aware DQN aggregator that outperforms random, linear-Q, policy gradient, and static robust aggregation baselines on robustness-accuracy trade-off

🛡️ Threat Analysis

Details

Similar Papers