Title: Secure exchange and incentivised collaboration of machine learning models and datasets using blockchain

Authors: Jessica Judith D'souza; R. Padmavathy; Abhinav Dayal; Harika Gnanaveni; Sreenu Ponnada

Addresses: Department of Computer Science, University of Southern California, CA, USA ' Department of Computer Science and Engineering, NIT Warangal, TS, India ' Department of Computer Science and Engineering, Vishnu Institute of Technology Bhimavaram, AP, India ' Department of Computer Science and Engineering, National Institute of Technology Warangal, TS, India ' Department of Computer Science and Engineering, Vishnu Institute of Technology Bhimavaram, AP, India

Abstract: Machine learning is an extremely centralised platform and foments a skewed dynamic in which only a few people can utilise the enormous opportunities it provides. There is an absolute necessity to build a decentralised platform where the users may exchange the trained models and datasets to solve machine learning problems collaboratively. In this paper, a secure way of exchanging these using a distributed trustless scalable platform through blockchain is proposed, thereby making the entire process seamless instead of relying on the traditional centralised platforms. Further, the performance is improved compared with other platforms by using two approaches such as parallelisation techniques and reducing gas usage. An authentication layer is added to improve the security of the whole system. The proposed design additionally incorporates an incentive mechanism to encourage more individuals to learn machine learning to contribute and hence, drawing them towards collaborative learning.

Keywords: blockchain; machine learning; collaboration; incentive mechanism; authentication.

DOI: 10.1504/IJICS.2022.127171

International Journal of Information and Computer Security, 2022 Vol.19 No.3/4, pp.443 - 462

Received: 22 Feb 2022
Accepted: 13 Jun 2022
Published online: 23 Nov 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article