面向资源最优分配的深度学习双边拍卖算法  

Deep learning double auction algorithm for resource optimal allocation

作　　者：郑阳超 李珍妮 ZHENG Yang-chao;LI Zhen-ni(School of Automation,Guangdong University of Technology,Guangzhou Guangdong 510006,China)

机构地区：[1]广东工业大学自动化学院,广东广州510006

出　　处：《控制理论与应用》2023年第10期1863-1872,共10页Control Theory & Applications

基　　金：广州市基础研究计划基础与应用基础研究项目(202002030289)资助。

摘　　要：针对拍卖过程中计算效率低和利益分配不合理等问题,本文提出了一种基于深度学习的迭代双边拍卖算法.该算法通过买卖双方的初始报价数据训练基于神经网络的资源最优分配模型,调用训练好的模型对实时报价数据快速响应,直接求解经纪人最优分配问题(BAP)以实现计算资源分配,显著地减小了计算代价,提高了算法的计算效率.进一步,针对利益分配不合理等问题,在迭代双边拍卖框架的支出规则和收入规则中引入调节因子用于调节买卖双方的利益,解决已有算法在实现社会福利最大化过程中利益分配不合理的问题.实验结果验证了该算法的有效性和优越性,在运行时间、社会福利、买家利益、卖家利益和经纪人利益等多项指标均明显优于已有的迭代双边拍卖算法.To solve the problems of low computational efficiency and unreasonable utilities distribution in the process of auction,this paper proposes an iterative double auction algorithm based on deep learning.The initial bidding data of both buyers and sellers are used to train resources optimal allocation model in the algorithm,then invoking the trained model quick responds to the real-time bidding data directly solving the broker optimal allocation problem(BAP)to reach computing resource optimal,which significantly reduces the calculation cost,consumes less time,and improves the efficiency of the algorithm.Further,in view of problems such as unreasonable distribution of utilities,adjustment factors are introduced into the spending rule and earning rule of the iterative double auction framework to adjust the utilities of the buyers and sellers,which solves the unreasonable utilities distribution in the process of maximizing social welfare under the existing algorithm.The experimental results show that the proposed algorithm is superior to the existing iterative double auction algorithm in terms of running time,social welfare,the utilities of buyers and sellers,the utilities of broker and other indicators.

关 键 词：深度学习 迭代双边拍卖 资源最优分配模型 调节因子 社会福利 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]