Journal article

How to Democratise and Protect AI: Fair and Differentially Private Decentralised Deep Learning

L Lyu, Y Li, K Nandakumar, J Yu, X Ma

IEEE Transactions on Dependable and Secure Computing | Published : 2020

Abstract

This paper firstly considers the research problem of fairness in collaborative deep learning, while ensuring privacy. We study the weaknesses in current server-based deep learning frameworks, and solve the single-point-of-failure problem by using Blockchain to realise decentralisation. To address fairness and privacy, we propose a novel reputation system through digital tokens and local credibility to ensure fairness, in combination with differential privacy to guarantee privacy. In particular, we build a fair and differentially private decentralised deep learning framework called FDPDDL, which enables parties to derive more accurate local models in a fair and private manner by using our dev..

View full abstract

University of Melbourne Researchers

Citation metrics