Poisoning Attacks and Defenses in Federated Learning: A Survey
This work identifies and analyzes poisoning risks in federated learning, which is crucial for developers and researchers in privacy-sensitive applications, but it is incremental as a survey.
The survey addresses security threats in federated learning by categorizing poisoning attacks and evaluating defenses, highlighting the need for robust systems due to privacy-preserving but vulnerable distributed training.
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.