CHEM: Efficient Secure Aggregation with Cached Homomorphic Encryption in Federated Machine Learning Systems

TL;DR

CHEM introduces a caching protocol for homomorphic encryption in federated learning, significantly reducing encryption costs while maintaining security, enabling more efficient privacy-preserving machine learning.

Contribution

The paper presents CHEM, a caching-based homomorphic encryption scheme that reduces computational overhead in secure federated learning without compromising security.

Findings

Reduces encryption costs by up to 89% in confidential inference.

Achieves sub-second overhead in practical scenarios.

Maintains semantic security under practical assumptions.

Abstract

Although homomorphic encryption can be incorporated into neural network layers for securing machine learning tasks, such as confidential inference over encrypted data samples and encrypted local models in federated learning, the computational overhead has been an Achilles heel. This paper proposes a caching protocol, namely CHEM, such that tensor ciphertexts can be constructed from a pool of cached radixes rather than carrying out expensive encryption operations. From a theoretical perspective, we demonstrate that CHEM is semantically secure and can be parameterized with straightforward analysis under practical assumptions. Experimental results on three popular public data sets show that adopting CHEM only incurs sub-second overhead and yet reduces the encryption cost by 48%--89% for encoding input data samples in confidential inference and 67%--87% for encoding local models in federated learning, respectively.