StratFL: New defense against model poisoning attacks in federated learning
In brief
- StratFL is an algorithm designed to counter model poisoning attacks in federated learning.
- It improves accuracy by 10% compared to existing methods in tests on MNIST and CIFAR-10.
- The approach models server-attacker interaction as an imperfect repeated game.
Researchers introduce StratFL, an algorithm to defend against model poisoning attacks in federated learning. StratFL improves accuracy by 10% over existing methods and models the server-attacker interaction as an imperfect repeated game.
What it changes for you
- HealthStratFL enhances security of federated models in healthcare, reducing risks of poisoning attacks.
- Finance & BankingStratFL mitigates vulnerabilities in federated learning systems used in finance, protecting sensitive data.
This summary gives you the essentials. The full article, its figures and quotes are on cs.AI updates on arXiv.org.
Read the original articleAI-generated summary of the source, checked by rules. Report an error