学术报告 | 大规模数据和模型市场的定价：模型、公平性与可扩展性

浙江大学网络空间安全学院

学术报告

  摘 要  

Data markets are emerging and promising for harvesting data from many data owners to support data-driven AI applications and many second-uses of big data. Pricing plays a central role in data markets.  In this talk, I will survey the motivations and the state-of-the-art practice of data and (machine learning) model markets, and review data pricing in end-to-end data analytics and machine learning pipelines.  Then, I will focus on models, fairness, and scalability of data pricing using some well established solution concepts in cooperative game theory, such as Shapley value. As a principled approach, I will illustrate that with some simple yet practical assumptions about the utility of data products, assessing exact Shapley value of millions of products and tens of owners is highly practical. I will also demonstrate the challenges in modeling and computing fair reward allocation in one-shot cooperative machine learning processes, such as federated learning, as well as in building privacy preserving model marketplaces.

数据市场正在兴起，并有望从许多数据所有者那里收集数据，以支持数据驱动的人工智能应用和大数据的许多二次使用。定价在数据市场中起着核心作用。在这个报告中，我将回顾数据和（机器学习）模型市场的动机和实践，并探索端到端数据分析和机器学习中的数据定价。然后，我将使用合作博弈论中一些成熟的解决方案概念（例如 Shapley 值）来关注数据定价的模型、公平性和可扩展性。在一些对数据产品效用的简单而实用的假设下，评估数百万个产品和数十个所有者的 Shapley 值是实际可行的。我还将分析在一次性协作机器学习过程（例如联邦学习）中建模和计算公平奖励分配的挑战，以及建立隐私保护模型市场方面的挑战。

  报告人简介  

Jian Pei is Professor at Duke University. His research focuses on data science, data mining, database systems, information retrieval and applied machine learning. His expertise is on developing effective and efficient data analysis techniques for novel data intensive applications, and transferring to products and business practice. He is recognized as a Fellow of the Royal Society of Canada (Canada's national academy), the Canadian Academy of Engineering, ACM and IEEE. He received several prestigious awards, including the 2017 ACM SIGKDD Innovation Award, the 2015 ACM SIGKDD Service Award, and the 2014 IEEE ICDM Research Contributions Award. He was a past chair of ACM SIGKDD and a past EIC of IEEE TKDE.

裴健博士是杜克大学教授。他的研究重点是数据科学、大数据、数据挖掘、数据库系统和信息检索。他的专长是为新型数据密集型应用开发高效灵活的数据分析技术，并将他的研究成果转化为行业产品和商业实践。他是加拿大皇家学会（即加拿大国家科学院）和加拿大工程院院士、ACM 和 IEEE 的会士。自2000年以来，他出版了一本教科书，两部专著，在学术期刊和会议上发表了300多篇研究论文，被广泛引用。他获得了一系列奖项，包括 2017 年 ACM SIGKDD 创新奖、2015 年 ACM SIGKDD 服务奖和 2014 年 IEEE ICDM 研究贡献奖。他是 ACM SIGKDD 的前任主席和 IEEE TKDE 的前任主编。