Incentive Mechanism Design for Federated Learning

In federated learning, motivating data owners to continue participating in a data federation is an important challenge. The key to achieving this objective is to device an incentive scheme that shares the profit generated by a federation with participants in a fair and just manner. Before this step can be achieved, a mechanism for evaluating the contribution toward the federated model by a given data owner must be established. Although there has yet to be published research on solving this problem, there has been a well-established line of work on using auction-based approaches to motivate sensors to commit more resources to improve data quality, which might shed light into solving this ...

Get Federated Learning now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.