Poisoning Attacks and Defenses in Federated Learning: A Survey

Abstract--Federated learning (FL) enables the training of models among distributed clients without compromising the privacy of training datasets, while the invisibility of clients' datasets and the training process poses a variety of security threats. This survey provides the taxonomy of poisoning attacks and experimental evaluation to discuss the need for robust FL. The FL has emerged as a promising solution to a authors also conducted an experimental evaluation in order number of applications to solve data silos while protecting to draw a conclusion on how to select the suitable method the privacy of data. Since its emergence, FL has been in each category of adversarial attacks. Furthermore, a brief employed in a variety of applications including but not limited overview of threats to FL is discussed in [2] and focuses on to healthcare, crowdsourcing systems, natural language poisoning and inference attacks in order to comprehend the processing (NLP), and the Internet of Things (IoT).