A Contribution Evaluation and Reward Framework Based on Shapley Value and Smart Contract

作者

  • Shengdong Chen Guangzhou Institute of Software 作者
  • Zhanwei Guo Zhongke Software Testing (Guangzhou) Co., Ltd. 作者
  • Tianyou Liu Guangzhou Zhiying Weilai Technology Co., Ltd. 作者
  • Tianwen Zhang Hebei Zhuowei 'er Information Technology Co., LTD 作者

關鍵詞:

Federal Learning, Incentive mechanism Shapley value, Smart contract Contribution Evaluation

摘要

Currently, due to the advantages of Shapley values in explaining model decisions, incentive mechanisms based on Shapley value contribution assessment have become a major research focus. This approach ensures the effectiveness and fairness of contribution assessment algorithms. However, there is a problem regarding the reliability of the computational results. Additionally, traditional Shapley value-based contribution processes face an issue where computational complexity increases exponentially with the number of participants. To address these issues, this paper proposes a contribution assessment and reward framework based on Shapley values and smart contract technology using alliance blockchain. This framework overcomes the reliance on third-party institutions that is characteristic of traditional incentive mechanisms, leveraging the decentralized nature of blockchain to eliminate trust issues associated with a single centralized entity. Furthermore, the traceability of blockchain ensures the transparency and traceability of the contribution assessment and reward distribution processes.

已發表

2024-06-01