NRT-DESE: Team Science for Integrative Graduate Training in Data Science and Physical Science

NRT-DESE：数据科学和物理科学研究生综合培训的团队科学

• 批准号: 1633631
• 负责人: Padhraic Smyth
• 金额: $ 296.72万
• 依托单位: University of California-Irvine
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-09-15 至 2021-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1633631&HistoricalAwards=false
• 关键词: NRT; DESE; Team; Science; Integrative

Massively parallel computers simulate data about molecular phenomena at previously unimaginable scales, satellites scan the planet capturing vast sets of measurements about ecosystem health, and particle accelerators generate tremendous amounts of data revealing fundamental properties of the smallest building blocks of matter; all with potentially broad societal benefits in areas such as drug discovery, energy conservation, and materials science. To fully realize these benefits will require a workforce with the technical skills to extract useful information from massive scientific data sets, calling for new approaches to graduate student training that emphasize expertise in data-driven science. This National Science Foundation Research Traineeship (NRT) award to the University of California Irvine (UCI) will tackle this challenge by creating a training ecosystem comprised of leading UCI, national-laboratory, and private-sector researchers across particle physics, earth science, chemistry, statistics and machine learning; all bound together by expertise in the emerging Science of Team Science. The project anticipates training over sixty (60) MS and PhD students, including twenty (20) funded trainees, from diverse backgrounds in computational statistics, machine learning, earth science, particle physics, synthetic chemistry, and team science. After graduation, students from this program will have both the technical and team-science skills to be leaders in the emerging field of data-driven science, and to participate in and lead interdisciplinary research teams at national laboratories, in academia, and in industry labs.The research agenda of the program seeks to create the foundation from which bridges can be built between the traditional scientific route of building interpretable models based on physical principles and data-driven modeling approaches that can provide high fidelity predictions but may lack clear interpretability in terms of the underlying science. The program will involve a number of interrelated research themes across multiple disciplines in the information and physical sciences, including machine learning (e.g. temporal and spatial data modeling, multi-scale models, deep learning, and scalable learning algorithms), particle and astroparticle physics (e.g. accelerator based experiments), earth systems science (e.g. reducing ecosystem response prediction uncertainties), and chemistry (e.g. prediction of physical properties of small molecules). A significant aspect of the program is an emphasis on team science as a core theme. Students will collaborate in small interdisciplinary research teams consisting of students and faculty with different disciplinary skills, and will take part in team-science workshops leading to student-led development of a team-science certificate in years 3 to 5 of the program. Summer internships for student participants, at both national and industry research laboratories, will serve to reinforce the students' academic training via participation in large-scale interdisciplinary data science research projects.The NSF Research Traineeship (NRT) Program is designed to encourage the development and implementation of bold, new potentially transformative models for STEM graduate education training. The Traineeship Track is dedicated to effective training of STEM graduate students in high priority interdisciplinary research areas, through the comprehensive traineeship model that is innovative, evidence-based, and aligned with changing workforce and research needs.

大规模并行计算机以以前无法想象的规模模拟有关分子现象的数据，卫星扫描地球捕获有关生态系统健康的大量测量数据，粒子加速器产生大量数据，揭示物质最小构建块的基本属性;所有这些都在药物发现，能源节约和材料科学等领域具有潜在的广泛社会效益。为了充分实现这些好处，需要一支具有从大量科学数据集中提取有用信息的技术技能的劳动力，这需要采用强调数据驱动科学专业知识的新研究生培训方法。这项授予加州尔湾大学（UCI）的国家科学基金会研究培训计划（NRT）将通过创建一个培训生态系统来应对这一挑战，该生态系统由领先的UCI、国家实验室和私营部门研究人员组成，涵盖粒子物理学、地球科学、化学、统计学和机器学习;所有这些都由新兴科学团队科学的专业知识结合在一起。该项目预计将培训六十（60）名硕士和博士生，其中包括二十（20）名受资助的学员，他们来自计算统计，机器学习，地球科学，粒子物理，合成化学和团队科学的不同背景。毕业后，该计划的学生将拥有技术和团队科学技能，成为数据驱动科学新兴领域的领导者，并参与和领导国家实验室，学术界，该计划的研究议程旨在建立基础，在构建可解释模型的传统科学路线之间建立桥梁，基于物理原理和数据驱动的建模方法，可以提供高保真度的预测，但可能缺乏明确的解释性的基础科学。该计划将涉及信息和物理科学多个学科的许多相互关联的研究主题，包括机器学习（例如时空数据建模、多尺度模型、深度学习和可扩展学习算法）、粒子和天体粒子物理学（例如基于加速器的实验）、地球系统科学（例如减少生态系统响应预测的不确定性）和化学（例如预测小分子的物理性质）。该计划的一个重要方面是强调团队科学作为核心主题。学生将在由具有不同学科技能的学生和教师组成的小型跨学科研究团队中进行合作，并将参加团队科学研讨会，从而在该计划的第3年至第5年以学生为主导开发团队科学证书。在国家和行业研究实验室为学生提供暑期实习，通过参与大规模跨学科数据科学研究项目，加强学生的学术培训。NSF研究培训（NRT）计划旨在鼓励开发和实施大胆的、新的潜在变革性的STEM研究生教育培训模式。该培训轨道致力于在高优先级的跨学科研究领域的STEM研究生的有效培训，通过全面的培训模式，是创新的，以证据为基础，并与不断变化的劳动力和研究需求保持一致。

项目成果

Multilevel Artificial Neural Network Training for Spatially Correlated Learning

• DOI: 10.1137/18m1191506
• 发表时间: 2018-06
• 期刊: SIAM J. Sci. Comput.
• 作者: C. Scott;E. Mjolsness

Forecasting Daily Wildfire Activity Using Poisson Regression

• DOI: 10.1109/tgrs.2020.2968029
• 发表时间: 2020-02
• 期刊: IEEE Transactions on Geoscience and Remote Sensing
• 影响因子: 8.2
• 作者: Casey A. Graff;S. Coffield;Yang Chen;E. Foufoula‐Georgiou;J. Randerson;Padhraic Smyth

Learning to isolate muons

学习分离μ子

• DOI: 10.1007/jhep10(2021)200
• 发表时间: 2021
• 期刊: Journal of High Energy Physics
• 影响因子: 5.4
• 作者: Collado, Julian;Bauer, Kevin;Witkowski, Edmund;Faucett, Taylor;Whiteson, Daniel;Baldi, Pierre
• 通讯作者: Baldi, Pierre

Generative Modeling of Atmospheric Convection

• DOI: 10.1145/3429309.3429324
• 发表时间: 2020-07
• 期刊: Proceedings of the 10th International Conference on Climate Informatics
• 作者: G. Mooers;Jens Tuyls;S. Mandt;M. Pritchard;T. Beucler

Mapping machine-learned physics into a human-readable space

将机器学习的物理映射到人类可读的空间

• DOI: 10.1103/physrevd.103.036020
• 发表时间: 2021
• 期刊: Physical Review D
• 影响因子: 5
• 作者: Faucett, Taylor;Thaler, Jesse;Whiteson, Daniel
• 通讯作者: Whiteson, Daniel