Deakin University
Browse

File(s) under permanent embargo

Blockchain-driven privacy-preserving machine learning

Version 2 2024-06-04, 04:01
Version 1 2021-09-17, 08:10
chapter
posted on 2024-06-04, 04:01 authored by Y Qu, Longxiang GaoLongxiang Gao, Yong XiangYong Xiang
With the integration of blockchain with current leading privacy-preserving machine learning mechanism, the performances of FL and GAN-DP can be further improved, especially the robustness against poisoning attacks. In addition, the deployment of blockchain as the underlying architecture enables decentralization while providing incentive mechanisms. Furthermore, the efficiency can be guaranteed, and the storage resources can be saved with an off-chain structure. Future directions in this field may include the optimization using game theory and reversible blockchain using chameleon hash. Chapter Contents: • 8.1 GAN-DP and blockchain • 8.1.1 Wasserstein generative adversarial net • 8.1.2 Generator and discriminator • 8.1.3 GAN-DP with a DP identifier • 8.1.4 Decentralized privacy • 8.1.5 Further discussion • 8.2 Federated learning and blockchain • 8.2.1 Existing issues • 8.2.2 How blockchain benefits FL • 8.2.3 Blockchain-enabled federated learning • 8.3 Conclusion remarks • References.

History

Chapter number

8

Pagination

189-200

ISBN-13

9781785618734

Language

eng

Publication classification

B1 Book chapter

Extent

12

Editor/Contributor(s)

Huang H, Wang L, Wu Y, Choo K-KR

Publisher

Institution of Engineering & Technology

Place of publication

London, Eng.

Title of book

Blockchains for network security: Principles, technologies and applications

Usage metrics

    Research Publications

    Categories

    No categories selected

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC