Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Efficient and Privacy-Preserving Federated Learning based on Full Homomorphic Encryption

Yuqi Guo, Lin Li, Zhongxiang Zheng, Hanrui Yun, Ruoyan Zhang, Xiaolin Chang, Zhixuan Gao

TL;DR

This work integrates fully homomorphic encryption (FHE), specifically the CKKS scheme, into both horizontal and vertical federated learning to enhance security while maintaining practical efficiency. It introduces a CKKS-based SecureBoost framework with approximate split finding, ciphertext packing, and a PSI-assisted federated inference protocol to reduce communication and protect privacy. It also extends FHE to logistic regression for both horizontal and vertical settings, including data binning (WOE) and SMOTE implemented in the encrypted domain, achieving significant speedups over traditional Paillier-based or non-FHE approaches. Across medical, business, biometric, and financial datasets, the results demonstrate improved training efficiency, reduced communication, and robust resistance to quantum and gradient-based attacks, highlighting the approach’s practical potential in privacy-preserving machine learning.

Abstract

Since the first theoretically feasible full homomorphic encryption (FHE) scheme was proposed in 2009, great progress has been achieved. These improvements have made FHE schemes come off the paper and become quite useful in solving some practical problems. In this paper, we propose a set of novel Federated Learning Schemes by utilizing the latest homomorphic encryption technologies, so as to improve the security, functionality and practicality at the same time. Comparisons have been given in four practical data sets separately from medical, business, biometric and financial fields, covering both horizontal and vertical federated learning scenarios. The experiment results show that our scheme achieves significant improvements in security, efficiency and practicality, compared with classical horizontal and vertical federated learning schemes.

Efficient and Privacy-Preserving Federated Learning based on Full Homomorphic Encryption

TL;DR

This work integrates fully homomorphic encryption (FHE), specifically the CKKS scheme, into both horizontal and vertical federated learning to enhance security while maintaining practical efficiency. It introduces a CKKS-based SecureBoost framework with approximate split finding, ciphertext packing, and a PSI-assisted federated inference protocol to reduce communication and protect privacy. It also extends FHE to logistic regression for both horizontal and vertical settings, including data binning (WOE) and SMOTE implemented in the encrypted domain, achieving significant speedups over traditional Paillier-based or non-FHE approaches. Across medical, business, biometric, and financial datasets, the results demonstrate improved training efficiency, reduced communication, and robust resistance to quantum and gradient-based attacks, highlighting the approach’s practical potential in privacy-preserving machine learning.

Abstract

Since the first theoretically feasible full homomorphic encryption (FHE) scheme was proposed in 2009, great progress has been achieved. These improvements have made FHE schemes come off the paper and become quite useful in solving some practical problems. In this paper, we propose a set of novel Federated Learning Schemes by utilizing the latest homomorphic encryption technologies, so as to improve the security, functionality and practicality at the same time. Comparisons have been given in four practical data sets separately from medical, business, biometric and financial fields, covering both horizontal and vertical federated learning scenarios. The experiment results show that our scheme achieves significant improvements in security, efficiency and practicality, compared with classical horizontal and vertical federated learning schemes.
Paper Structure (35 sections, 62 equations, 17 figures, 11 tables)

This paper contains 35 sections, 62 equations, 17 figures, 11 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Notations
  4. Federated Learning
  5. Full Homomorphic Encryption
  6. Private Set Intersection
  7. Federated Learning with SecureBoost based on FHE
  8. Classical SecureBoost Model based on PHE
  9. XGBoost Model
  10. SecureBoost Model
  11. Federated Learning with SecureBoost based on FHE
  12. Approximate Algorithm for Split Finding
  13. Ciphertext packing
  14. Subtraction between tree nodes
  15. Federated Inference based on the Learned Model
  16. ...and 20 more sections

Figures (17)

  • Figure 1: Horizontal federated learning framework.
  • Figure 2: Vertical federated learning framework.
  • Figure 3: Schematic diagram for finding candidate split points.
  • Figure 4: Subtraction between tree nodes.
  • Figure 5: An illustration of classic federated inference.
  • ...and 12 more figures