Confidential Federated Learning for Heterogeneous Platforms against Client-Side Privacy Leakages

Federated learning can mitigate privacy concerns. However, it remains vulnerable to privacy breaches at both the server aggregation and client training ends. Currently, enhancing privacy at the server aggregation side is a prominent focus in cutting-edge federated learning research. Nonetheless, threats originating from clients, particularly those posed by untrusted clients to the privacy of other clients, are gaining increasing attention. This work presents a privacy protection framework for federated learning under client-originated threats. The framework is compatible with heterogeneous platforms, integrating the confidentiality of Trusted Execution Environments (TEE) with the high performance of GPUs.