Quantum delegated and federated learning via quantum homomorphic encryption

TL;DR

This work presents a general framework that enables quantum delegated and federated learning with a computation-theoretical data privacy guarantee, and shows that learning and inference under this framework feature substantially lower communication complexity compared with schemes based on blind quantum computing.

Abstract

Quantum learning models hold the potential to bring computational advantages over the classical realm. As powerful quantum servers become available on the cloud, ensuring the protection of clients' private data becomes crucial. By incorporating quantum homomorphic encryption schemes, we present a general framework that enables quantum delegated and federated learning with a computation-theoretical data privacy guarantee. We show that learning and inference under this framework feature substantially lower communication complexity compared with schemes based on blind quantum computing. In addition, in the proposed quantum federated learning scenario, there is less computational burden on local quantum devices from the client side, since the server can operate on encrypted quantum data without extracting any information. We further prove that certain quantum speedups in supervised learning carry over to private delegated learning scenarios employing quantum kernel methods. Our results provide a valuable guide toward privacy-guaranteed quantum learning on the cloud, which may benefit future studies and security-related applications.

Figures (2)

• Figure 1: A schematic illustration of quantum delegated and federated learning adapting quantum homomorphic encryption techniques. On the left side, we exhibit single-client quantum delegated learning. For the training data in the form of quantum states or classical bits, the client applies a quantum or classical one-time pad to encrypt it, respectively. Upon receiving the data, the server homomorphically operates on the encrypted data and returns the encrypted results, which contain the information for model optimization, to the client. After decrypting the results, the client could then update the model parameters. On the right side, the protocol is extended to the multi-party federated learning scenario, where different clients, each holding their private data, can collaboratively train a shared model.
• Figure : Quantum federated learning

Theorems & Definitions (1)

• Definition 1