Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SoK: Privacy Preserving Machine Learning using Functional Encryption: Opportunities and Challenges

Prajwal Panzade, Daniel Takabi

TL;DR

This SoK surveys privacy-preserving ML based on Functional Encryption (FE), distinguishing IPFE and QFE as the two primary FE families used for ML on encrypted data. It reviews core FE foundations, neural and polynomial neural networks, and analyzes two FE libraries (IPFE/ QFE) along with supporting cryptographic toolkits, dataset usage, and security notions. The paper highlights significant efficiency and usability challenges, including training-over-encrypted-data, limited FE functionality, and reliance on CPU implementations, while outlining opportunities such as multi-authority FE, open-source libraries, and hardware acceleration. Overall, FE-based PPML offers plaintext results from encrypted inputs, enabling new privacy-preserving ML workflows, but practical deployment requires advances in functionality, security, and performance.

Abstract

With the advent of functional encryption, new possibilities for computation on encrypted data have arisen. Functional Encryption enables data owners to grant third-party access to perform specified computations without disclosing their inputs. It also provides computation results in plain, unlike Fully Homomorphic Encryption. The ubiquitousness of machine learning has led to the collection of massive private data in the cloud computing environment. This raises potential privacy issues and the need for more private and secure computing solutions. Numerous efforts have been made in privacy-preserving machine learning (PPML) to address security and privacy concerns. There are approaches based on fully homomorphic encryption (FHE), secure multiparty computation (SMC), and, more recently, functional encryption (FE). However, FE-based PPML is still in its infancy and has not yet gotten much attention compared to FHE-based PPML approaches. In this paper, we provide a systematization of PPML works based on FE summarizing state-of-the-art in the literature. We focus on Inner-product-FE and Quadratic-FE-based machine learning models for the PPML applications. We analyze the performance and usability of the available FE libraries and their applications to PPML. We also discuss potential directions for FE-based PPML approaches. To the best of our knowledge, this is the first work to systematize FE-based PPML approaches.

SoK: Privacy Preserving Machine Learning using Functional Encryption: Opportunities and Challenges

TL;DR

This SoK surveys privacy-preserving ML based on Functional Encryption (FE), distinguishing IPFE and QFE as the two primary FE families used for ML on encrypted data. It reviews core FE foundations, neural and polynomial neural networks, and analyzes two FE libraries (IPFE/ QFE) along with supporting cryptographic toolkits, dataset usage, and security notions. The paper highlights significant efficiency and usability challenges, including training-over-encrypted-data, limited FE functionality, and reliance on CPU implementations, while outlining opportunities such as multi-authority FE, open-source libraries, and hardware acceleration. Overall, FE-based PPML offers plaintext results from encrypted inputs, enabling new privacy-preserving ML workflows, but practical deployment requires advances in functionality, security, and performance.

Abstract

With the advent of functional encryption, new possibilities for computation on encrypted data have arisen. Functional Encryption enables data owners to grant third-party access to perform specified computations without disclosing their inputs. It also provides computation results in plain, unlike Fully Homomorphic Encryption. The ubiquitousness of machine learning has led to the collection of massive private data in the cloud computing environment. This raises potential privacy issues and the need for more private and secure computing solutions. Numerous efforts have been made in privacy-preserving machine learning (PPML) to address security and privacy concerns. There are approaches based on fully homomorphic encryption (FHE), secure multiparty computation (SMC), and, more recently, functional encryption (FE). However, FE-based PPML is still in its infancy and has not yet gotten much attention compared to FHE-based PPML approaches. In this paper, we provide a systematization of PPML works based on FE summarizing state-of-the-art in the literature. We focus on Inner-product-FE and Quadratic-FE-based machine learning models for the PPML applications. We analyze the performance and usability of the available FE libraries and their applications to PPML. We also discuss potential directions for FE-based PPML approaches. To the best of our knowledge, this is the first work to systematize FE-based PPML approaches.
Paper Structure (44 sections, 4 equations, 3 figures, 4 tables)

This paper contains 44 sections, 4 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Background Knowledge
  3. Inner-product Functional Encryption
  4. Quadratic Functional Encryption
  5. Neural networks
  6. Polynomial Neural Network
  7. Scope and Methodology
  8. Overview of FE-based PPML models and libraries
  9. What has been done?
  10. Inner-product FE-based Machine Learning
  11. Quadratic FE-based Machine Learning
  12. Cryptography libraries for Functional Encryption implementation
  13. CiFEr
  14. GoFE
  15. FLINT
  16. ...and 29 more sections

Figures (3)

  • Figure 1: Overview of PPML using Functional Encryption
  • Figure 2: IPFE-based PPML
  • Figure 3: QFE-based PPML