Mitigating Leakage in Federated Learning with Trusted Hardware
This work addresses privacy risks for parties in federated learning, but it is incremental as it builds upon and improves the existing SecureBoost method.
The paper tackles the problem of information leakage in federated learning, specifically in the SecureBoost framework, by demonstrating a leakage-abuse attack and proposing two secure versions using trusted execution environments, resulting in protocols that are 1.2-5.4 times faster in computation and require 5-49 times less communication than SecureBoost.
In federated learning, multiple parties collaborate in order to train a global model over their respective datasets. Even though cryptographic primitives (e.g., homomorphic encryption) can help achieve data privacy in this setting, some partial information may still be leaked across parties if this is done non-judiciously. In this work, we study the federated learning framework of SecureBoost [Cheng et al., FL@IJCAI'19] as a specific such example, demonstrate a leakage-abuse attack based on its leakage profile, and experimentally evaluate the effectiveness of our attack. We then propose two secure versions relying on trusted execution environments. We implement and benchmark our protocols to demonstrate that they are 1.2-5.4X faster in computation and need 5-49X less communication than SecureBoost.