Privacy Preserved Decentralized Deep Learning: A Blockchain Based Solution for Secure AI-Driven Enterprise

Abstract

Deep learning and Blockchain attracted the attention of both the research community and the industry. In the financial enterprise by using the Blockchain technology, financial transactions could be performed in shorter periods and with higher transparency and security. In Blockchain ecosystem, there is no need for having a central reliable authority to regulate and control the system. In Blockchain many entities which cannot trust each other in normal conditions can join together to achieve a mutual goal. Deep learning algorithms are currently the best solution for many machine learning applications and provide high accuracy models for robotics, computer vision, smart cities and other AI-driven enterprise. However, availability of more data can boost the performance of deep models considerably. In this paper, a secure decentralized deep learning framework for big data analytics on Blockchain for AI-driven enterprise is proposed. The proposed framework uses the Stellar Blockchain infrastructure for secure decentralized training of the deep models. A Deep Learning Coin (DLC) is used for Blockchain compensation. The security of the proposed framework incentivizes people and organizations to share their valuable data for training the deep neural models while the privacy of their data is preserved.