Detection and Prevention Against Poisoning Attacks in Federated Learning
This addresses security vulnerabilities in federated learning systems, but it is incremental as it builds on existing detection techniques.
The paper tackles the problem of poisoning attacks in federated learning by proposing an average accuracy deviation detection (AADD) method to identify and blacklist malicious clients, showing promising results in preventing global model accuracy deterioration.
This paper proposes and investigates a new approach for detecting and preventing several different types of poisoning attacks from affecting a centralized Federated Learning model via average accuracy deviation detection (AADD). By comparing each client's accuracy to all clients' average accuracy, AADD detect clients with an accuracy deviation. The implementation is further able to blacklist clients that are considered poisoned, securing the global model from being affected by the poisoned nodes. The proposed implementation shows promising results in detecting poisoned clients and preventing the global model's accuracy from deteriorating.