Incentive-Based Ledger Protocols for Solving Machine Learning Tasks and Optimization Problems via Competitions

David Amar, Lior Zilpa; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2019, pp. 0-0

We propose incentive-based protocols that use competitions and public ledgers to solve optimization problems. We introduce Proof-of-Accumulated-Work (PoAW): miners compete in costumer-submitted jobs and accumulate recorded work on which they are later remunerated. These new competitions replace the standard hash puzzle-based competitions. A competition is managed by a dynamically-created small masternode network (dTMN) of invested miners, which improves scalability as we do not need the entire network to manage the competition. Using a careful design of incentives, our system preserves security, avoids attacks, and offers new markets to the miners. Finally, we illustrate how the new protocols can be used for implementing machine learning competitions.