Can I read Detection and Prevention Against Poisoning Attacks in Federated Learning on EtoBox?

Detection and Prevention Against Poisoning Attacks in Federated Learning by Valadi, Viktor; Englund, Madeleine; Spanier, Mark; O'brien, Austin is a scholarly article available to read on EtoBox.

What is Detection and Prevention Against Poisoning Attacks in Federated Learning about?

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.

Author

Valadi, Viktor; Englund, Madeleine; Spanier, Mark; O'brien, Austin

Published

2022

Language

EN