Towards Quantum-Safe Federated Learning via Homomorphic Encryption: Learning with Gradients

TL;DR

This paper proposes a novel homomorphic encryption framework for federated learning that preserves privacy and eliminates error expansion, enabling secure collaborative deep learning over untrusted servers.

Contribution

It introduces an error-free, LWE-based homomorphic encryption scheme tailored for federated learning, enhancing security and efficiency.

Findings

Eliminates error expansion in homomorphic encryption for gradients

Supports large-scale collaborative deep learning securely

Ensures cryptographic security of participant gradients

Abstract

This paper introduces a privacy-preserving distributed learning framework via private-key homomorphic encryption. Thanks to the randomness of the quantization of gradients, our learning with error (LWE) based encryption can eliminate the error terms, thus avoiding the issue of error expansion in conventional LWE-based homomorphic encryption. The proposed system allows a large number of learning participants to engage in neural network-based deep learning collaboratively over an honest-but-curious server, while ensuring the cryptographic security of participants' uploaded gradients.