Building Trusted Federated Learning: Key Technologies and Challenges

Federated learning (FL) provides convenience for cross-domain machine applications and has been widely studied. However, the original FL is still vulnerable to poisoning inference attacks, which will hinder landing application of FL. Therefore, it essential design a trustworthy federation (TFL) eliminate users’ anxiety. In this paper, we aim provide well-researched picture security privacy issues in that can bridge gap TFL. Firstly, define desired goals critical requirements TFL, observe model from perspective adversaries extrapolate roles capabilities potential backward. Subsequently, summarize current mainstream attack defense means analyze characteristics different methods. Based on priori knowledge, propose directions realizing future TFL deserve attention.