联邦学习中基于Chebyshev定理的模型性能感知逆向拍卖  

Model performance-aware reverse auction based on Chebyshev’s theorem in federated learning

作　　者：罗丰 王琦[1] 王青山[1] LUO Feng;WANG Qi;WANG Qingshan(School of Mathematics,Hefei University of Technology,Hefei 230601,China)

机构地区：[1]合肥工业大学数学学院,安徽合肥230601

出　　处：《合肥工业大学学报（自然科学版）》2024年第11期1486-1492,共7页Journal of Hefei University of Technology：Natural Science

基　　金：安徽省自然科学基金资助项目(2208085MF165)。

摘　　要：文章研究多服务器、多客户端联邦学习(federated learning,FL)场景中的激励机制,并将任务分配和定价问题建模为多个逆向拍卖问题。根据切比雪夫(Chebyshev)定理对客户端每一轮的本地模型性能进行评估,并进一步利用指数衰减函数评估其本地模型的总体性能;设计基于本地模型性能的逆向拍卖(local model performance based reverse auction,LPRA)算法解决任务分配和定价问题以激励更多高性能的客户端参与,并从理论上证明LPRA算法满足个体理性、真实性和计算高效性;通过仿真实验验证LPRA算法的有效性。This paper studies the incentive mechanism in a multi-server,multi-client federated learning(FL)scenario and models the task allocation and pricing as multiple reverse auction problem.Firstly,local model performance of clients is evaluated in each round according to Chebyshev’s theorem,and exponential decay function is used to evaluate the historical performance of clients.Then,a local model performance based reverse auction(LPRA)algorithm is designed to solve the task allocation and pricing problems with the goal of maximizing the overall performance of clients participating in FL.Through theoretical analysis,it is confirmed that LPRA algorithm satisfies individual rationality,truthfulness and computational efficiency.Finally,the effectiveness of the LPRA algorithm is verified by simulated experiments.

关 键 词：联邦学习(FL) 激励机制 切比雪夫定理 逆向拍卖 个体理性 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]