Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Auditable Homomorphic-based Decentralized Collaborative AI with Attribute-based Differential Privacy

Lo-Yao Yeh, Sheng-Po Tseng, Chia-Hsun Lu, Chih-Ya Shen

TL;DR

This work proposes a novel decentralized collaborative AI framework, named Auditable Homomorphic-based Decentralised Collaborative AI (AerisAI), to improve security with homomorphic encryption and fine-grained differential privacy and conducts extensive experiments on real datasets to evaluate the proposed approach.

Abstract

In recent years, the notion of federated learning (FL) has led to the new paradigm of distributed artificial intelligence (AI) with privacy preservation. However, most current FL systems suffer from data privacy issues due to the requirement of a trusted third party. Although some previous works introduce differential privacy to protect the data, however, it may also significantly deteriorate the model performance. To address these issues, we propose a novel decentralized collaborative AI framework, named Auditable Homomorphic-based Decentralised Collaborative AI (AerisAI), to improve security with homomorphic encryption and fine-grained differential privacy. Our proposed AerisAI directly aggregates the encrypted parameters with a blockchain-based smart contract to get rid of the need of a trusted third party. We also propose a brand-new concept for eliminating the negative impacts of differential privacy for model performance. Moreover, the proposed AerisAI also provides the broadcast-aware group key management based on ciphertext-policy attribute-based encryption (CPABE) to achieve fine-grained access control based on different service-level agreements. We provide a formal theoretical analysis of the proposed AerisAI as well as the functionality comparison with the other baselines. We also conduct extensive experiments on real datasets to evaluate the proposed approach. The experimental results indicate that our proposed AerisAI significantly outperforms the other state-of-the-art baselines.

Auditable Homomorphic-based Decentralized Collaborative AI with Attribute-based Differential Privacy

TL;DR

This work proposes a novel decentralized collaborative AI framework, named Auditable Homomorphic-based Decentralised Collaborative AI (AerisAI), to improve security with homomorphic encryption and fine-grained differential privacy and conducts extensive experiments on real datasets to evaluate the proposed approach.

Abstract

In recent years, the notion of federated learning (FL) has led to the new paradigm of distributed artificial intelligence (AI) with privacy preservation. However, most current FL systems suffer from data privacy issues due to the requirement of a trusted third party. Although some previous works introduce differential privacy to protect the data, however, it may also significantly deteriorate the model performance. To address these issues, we propose a novel decentralized collaborative AI framework, named Auditable Homomorphic-based Decentralised Collaborative AI (AerisAI), to improve security with homomorphic encryption and fine-grained differential privacy. Our proposed AerisAI directly aggregates the encrypted parameters with a blockchain-based smart contract to get rid of the need of a trusted third party. We also propose a brand-new concept for eliminating the negative impacts of differential privacy for model performance. Moreover, the proposed AerisAI also provides the broadcast-aware group key management based on ciphertext-policy attribute-based encryption (CPABE) to achieve fine-grained access control based on different service-level agreements. We provide a formal theoretical analysis of the proposed AerisAI as well as the functionality comparison with the other baselines. We also conduct extensive experiments on real datasets to evaluate the proposed approach. The experimental results indicate that our proposed AerisAI significantly outperforms the other state-of-the-art baselines.
Paper Structure (19 sections, 4 theorems, 9 figures, 1 table)

This paper contains 19 sections, 4 theorems, 9 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Works
  3. System Model and Assumptions
  4. The Proposed Scheme -- AerisAI
  5. Preliminaries
  6. Bilinear Pairing
  7. Ciphertext Policy Attribute-Based Encryption (CP-ABE)
  8. Paillier Encryption
  9. System Overview
  10. Workflow
  11. Security Analysis
  12. Evaluation
  13. Functionality Comparisons
  14. Performance Evaluation
  15. Experimental Settings
  16. ...and 4 more sections

Key Result

Theorem 1

Based on the difficulty of factoring large integers lenstra2000integer, our proposed AerisAI maintains the privacy that the encrypted gradients and noise cannot be decrypted without the corresponding private key.

Figures (9)

  • Figure 1: Overview of the proposed AriesAI.
  • Figure 2: Accuracy of different numbers of clients for MNIST.
  • Figure 3: Accuracy of different numbers of clients for CIFAR-10.
  • Figure 4: Accuracy of different numbers of clients for CIFAR-100.
  • Figure 5: Accuracy of different methods for MNIST, CIFAR-10, and CIFAR-100.
  • ...and 4 more figures

Theorems & Definitions (8)

  • Theorem 1
  • proof
  • Theorem 2
  • proof
  • Theorem 3
  • proof
  • Theorem 4
  • proof