MaSS：基于单位数据贡献的模型定价框架

doi:10.3778/j.issn.1673-9418.2207106

摘要/Abstract

摘要： 数据驱动型的机器学习模型已经是大势所趋，但是对于如何促进机器学习模型市场发展的研究还为之甚少。现有的研究主要分为两方面：一方面是数据拥有者和中间商之间的交互，即数据所有者的补偿问题；另一方面是模型买家与中间商之间的交互，即模型的定价问题。但是对于模型交易市场而言，这两个问题是密不可分的。因此针对这个问题，提出了新的模型定价框架MaSS。MaSS采用了新的效用评估指标Shapley值，然后基于该指标提出了三阶段的Stackelberg博弈的模型交易框架，其中包含了三方参与人：模型买家、中间人、数据拥有者。数据拥有者提供带有噪音的数据给中间人。中间人进一步将数据加工成模型买家需要的模型，向模型买家提供模型从而获利。他们之间相互作用形成一个Stackelberg博弈，以使得参与交易的所有人实现收益最大化，并从理论上证明了均衡定价策略存在的唯一性。最后，通过对真实数据大量模拟，证明了其显著的性能。

关键词: 数据定价, 模型定价, Shapley值, Stackelberg博弈

Abstract: Data-driven machine learning models have become ubiquitous. However, there is still very little research on how to promote the market development of the machine learning model. The existing research is mainly divided into two aspects, one is the interaction between the data owners and the brokers, that is, the compensation of the data owners. Another is the interaction between model buyers and brokers, that is, the expense of the model buyers. But for the model market, these issues are indivisible. Therefore, this paper takes a formal data marketplace perspective and proposes the novel model marketplace based on three-stage hierarchical Stackelberg game and Shapley value (MaSS). MaSS adopts a new utility evaluation index, Shapley value. And then this paper proposes a model trading framework of three-stage Stackelberg game based on Shapley value, including three-party parti-cipants: model buyers, brokers and data owners. The data owners provide the broker with private data. The brokers will further process the data into the models needed by the model buyers, and provide the model for the model buyer for profit. They interact with each other to form a Stackelberg game to maximize the profits of all involved in the transaction. And the uniqueness of the existence of equilibrium pricing strategy is proven theoretically. Finally, its remarkable performance is demonstrated by extensive simulations on real data.

Key words: data pricing, model pricing, Shapley value, Stackelberg game

张小伟, 江东, 袁野, 安丽霞, 王国仁. MaSS：基于单位数据贡献的模型定价框架[J]. 计算机科学与探索, 2023, 17(9): 2252-2264.

ZHANG Xiaowei, JIANG Dong, YUAN Ye, AN Lixia, WANG Guoren. MaSS: Model Pricing Marketplace Based on Unit Data Contribution[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(9): 2252-2264.

参考文献

[1] CHEN L, KOUTRIS P, KUMAR A. Towards model-based pricing for machine learning in a data marketplace[C]//Proceedings of the 2019 International Conference on Management of Data, Amsterdam, Jun 30-Jul 5, 2019. New York: ACM, 2019: 1535-1552.
[2] JIA R, DAO D, WANG B, et al. Efficient task-specific data valuation for nearest neighbor algorithms[J]. arXiv:1908. 08619, 2019.
[3] LIU J F, LOU J, LIU J X, et al. Dealer: an end-to-end model marketplace with differential privacy[J]. Proceedings of the VLDB Endowment, 2021, 14(6): 957-969.
[4] CASTRO J, GóMEZ D, TEJADA J. Polynomial calculation of the Shapley value based on sampling[J]. Computers & Operations Research, 2009, 36(5): 1726-1730.
[5] FATIMA S S, WOOLDRIDGE M, JENNINGS N R. A linear approximation method for the Shapley value[J]. Artificial Intelligence, 2008, 172(14): 1673-1699.
[6] GHORBANI A, ZOU J. Data Shapley: equitable valuation of data for machine learning[C]//Proceedings of the 36th International Conference on Machine Learning, Long Beach, Jun 9-15, 2019: 2242-2251.
[7] DWORK C, MCSHERRY F, NISSIM K, et al. Calibrating noise to sensitivity in private data analysis[C]//LNCS 3876:Proceedings of the Theory of Cryptography Conference,New York, Mar 4-7, 2006. Berlin, Heidelberg: Springer, 2006: 265-284.
[8] DWORK C, ROTH A. The algorithmic foundations of diffe-rential privacy[J]. Foundations and Trends in Theoretical Computer Science, 2014, 9(3/4): 211-407.
[9] AN B, XIAO M, LIU A, et al. Crowdsensing data trading based on combinatorial multi-armed bandit and Stackelberg game[C]//Proceedings of the 37th IEEE International Conf-erence on Data Engineering, Chania, Apr 19-22, 2021. Pisc-ataway: IEEE, 2021: 253-264.
[10] GU B, YANG X, LIN Z, et al. Multiagent actor-critic network-based incentive mechanism for mobile crowdsensing in industrial systems[J]. IEEE Transactions on Industrial Infor-matics, 2020, 17(9): 6182-6191.
[11] DUAN X, ZHAO C, HE S, et al. Distributed algorithms to compute Walrasian equilibrium in mobile crowdsensing[J]. IEEE Transactions on Industrial Electronics, 2016, 64(5): 4048-4057.
[12] ZHAN Y, LIU C H, ZHAO Y, et al. Free market of multi-leader multi-follower mobile crowdsensing: an incentive mec-hanism design by deep reinforcement learning[J]. IEEE Transactions on Mobile Computing, 2019, 19(10): 2316-2329.
[13] DUA D, GRAFF C. UCI machine learning repository[EB/OL]. [2022-05-09]. https://archive.ics.uci.edu/ml/datasets/ILPD+% 28Indian+Liver+Patient+Dataset%29.
[14] OLIVEIRA A C, FETZER C, MARTIN A, et al. Optimizing query prices for data-as-a-service[C]//Proceedings of the 2015 IEEE International Congress on Big Data, New York, Jun 27-Jul 2, 2015. Washington: IEEE Computer Society, 2015: 289-296.
[15] LUONG N C, HOANG D T, WANG P, et al. Data collec-tion and wireless communication in Internet of things (IoT) using economic analysis and pricing models: a survey[J]. IEEE Communications Surveys & Tutorials, 2016, 18(4): 2546-2590.
[16] LYU X Y, ZHANG R T, YUE J. Competition and coop-eration between participants of the Internet of things industry value chain[J]. Advances in Information Sciences & Service Sciences, 2012, 4(11): 406-412.
[17] MEI L J, LI W, NIE K. Pricing decision analysis for information services of the Internet of things based on Stackelberg game[C]//Proceedings of the 2nd International Conference on Logistics, Informatics and Service Science. Heidelberg: Springer, 2013: 1097-1104.
[18] JUNG K, PARK S. Privacy bargaining with fairness: privacy- price negotiation system for applying differential privacy in data market environments[C]//Proceedings of the 2019 IEEE International Conference on Big Data, Los Angeles, Dec 9-12, 2019. Piscataway: IEEE, 2019: 1389-1394.
[19] JANG B, PARK S, LEE J, et al. Three hierarchical levels of big-data market model over multiple data sources for Inter-net of things[J]. IEEE Access, 2018, 6: 31269-31280.
[20] 张小伟, 江东, 袁野. 基于博弈论和拍卖的数据定价综述[J]. 大数据, 2021, 7(4): 61-79.
ZHANG X W, JIANG D, YUAN Y. A survey of game theory and auction-based data pricing[J]. Big Data Research, 2021, 7(4): 61-79.
[21] KOUTRIS P, UPADHYAYA P, BALAZINSKA M, et al. Query-based data pricing[J]. Journal of the ACM, 2015, 62(5): 1-44.
[22] LIN B R, KIFER D. On arbitrage-free pricing for general data queries[J]. Proceedings of the VLDB Endowment, 2014, 7(9): 757-768.
[23] CHAWLA S, DEEP S, KOUTRIS P, et al. Revenue maxim-ization for query pricing[J]. arXiv:1909.00845, 2019.