TRIPODS: Institute for Foundations of Data Science

TRIPODS：数据科学研究所

• 批准号: 1740707
• 负责人: Stephen Wright
• 金额: $ 149.95万
• 依托单位: University of Wisconsin-Madison
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-09-01 至 2021-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1740707&HistoricalAwards=false
• 关键词: TRIPODS; Institute; Foundations; Data; Science

As data continues to accumulate at an ever-increasing rate, so does the need for powerful and novel methods to extract information from data, in a form that is useful to individuals, society, researchers,and commerce. This project establishes a new entity at the University of Wisconsin-Madison: the Institute for Foundations of Data Science (IFDS). Building on the foundational work of earlier generations of researchers, IFDS will serve as a hub for people across campus with expertise in mathematics, statistics, and computer science to explore new approaches to the formulation and solution of problems in data analysis, as well as to epitomize the possibilities of a collaborative approach to investigating fundamental issues in data science. IFDS will integrate with the broader UW-Madison agenda for data science research, creating a new home for research of a fundamental, theoretical nature. It will play a vital role in establishing graduate degree programs in data science and in outreach to industrial partners with interests in fundamental data science research.The new Institute for Foundations of Data Science will bring together researchers across the UW-Madison campus in mathematics, statistics, and theoretical computer science for a transdisciplinary effort organized around three themes: Algebra and Optimization in Data Science, Graphs and Networks in Data Science, and Data Acquisition Theory and Methods. All topics represent areas of significant current interest in data science due to their intrinsic fundamental import and their wide applicability. The collaborations within these themes will involve fourteen senior researchers, together with postdoctoral and graduate student researchers. The IFDS will lay the foundations for a larger future effort in transdisciplinary data science research, possibly involving other universities and institutes. Funds for the project come from CISE Computing and Communications Foundations, MPS Division of Mathematical Sciences, MPS Office of Multidisciplinary Activities, and Growing Convergent Research. (Convergence can be characterized as the deep integration of knowledge, techniques, and expertise from multiple fields to form new and expanded frameworks for addressing scientific and societal challenges and opportunities. This project promotes convergence by developing transdisciplinary data science solutions to problems in important application domains such as functional brain networks in cognitive neuroscience, understanding causal effects in gene regulatory networks, phylogenetic reconstruction in computational evolutionary biology, and the spread of information or disease across a network.)

随着数据以不断增长的速度不断积累，对从数据中提取信息的强大而新颖的方法的需求也越来越大，这些方法对个人，社会，研究人员和商业都有用。该项目在威斯康星大学麦迪逊分校建立了一个新的实体：数据科学基础研究所（IFDS）。在前几代研究人员的基础工作的基础上，IFDS将作为整个校园的人在数学，统计学和计算机科学方面的专业知识的中心，探索制定和解决数据分析问题的新方法，以及集中体现合作方法的可能性，以调查数据科学中的基本问题。IFDS将与更广泛的UW-Madison数据科学研究议程相结合，为基础理论研究创造一个新的家园。 它将在建立数据科学研究生学位课程以及与对基础数据科学研究感兴趣的行业合作伙伴进行外联方面发挥至关重要的作用。新的数据科学基础研究所将汇集威斯康星大学麦迪逊分校校园内数学，统计学和理论计算机科学的研究人员，围绕三个主题进行跨学科的努力：数据科学中的代数和优化，数据科学中的图形和网络，以及数据采集理论和方法。由于其内在的基本意义和广泛的适用性，所有主题都代表了数据科学中当前重要的领域。这些主题内的合作将涉及14名高级研究人员，以及博士后和研究生研究人员。IFDS将为未来跨学科数据科学研究的更大努力奠定基础，可能涉及其他大学和研究所。 该项目的资金来自CISE计算和通信基金会，MPS数学科学部，MPS多学科活动办公室和不断发展的融合研究。（融合可以被描述为来自多个领域的知识，技术和专业知识的深度整合，以形成新的和扩展的框架，以应对科学和社会的挑战和机遇。该项目通过开发跨学科的数据科学解决方案来促进融合，以解决重要应用领域的问题，如认知神经科学中的功能性脑网络，理解基因调控网络中的因果效应，计算进化生物学中的系统发育重建，以及信息或疾病在网络中的传播。

项目成果

Learning to Solve Linear Inverse Problems in Imaging with Neumann Networks

学习使用诺伊曼网络解决成像中的线性逆问题

• 发表时间: 2019
• 期刊: NeurIPS 2019 Workshop on Solving Inverse Problems with Deep Networks
• 作者: Ongie, Greg;Gilton, Davis;Willett, Rebecca
• 通讯作者: Willett, Rebecca

A Function Space View of Bounded Norm Infinite Width ReLU Nets: The Multivariate Case

• 发表时间: 2019-10
• 期刊: ArXiv
• 作者: Greg Ongie;R. Willett;Daniel Soudry;N. Srebro

Randomized sampling for basis functions construction in generalized finite element methods

广义有限元方法中基函数构造的随机采样

• 发表时间: 2020
• 期刊: Multiscale modeling simulation
• 作者: Chen, K;and Wright, S.J.
• 通讯作者: and Wright, S.J.

N-Gram Graph: Simple Unsupervised Representation for Graphs, with Applications to Molecules

• 发表时间: 2018-06
• 作者: Shengchao Liu;M. F. Demirel;Yingyu Liang

Stability and Generalization of Learning Algorithms that Converge to Global Optima

• 发表时间: 2017-10
• 作者: Zachary B. Charles;Dimitris Papailiopoulos