Collaborative Research: CyberTraining: Implementation: Medium: Cyber2A: CyberTraining on AI-driven Analytics for Next Generation Arctic Scientists

合作研究：网络培训：实施：媒介：Cyber​​2A：下一代北极科学家人工智能驱动分析的网络培训

• 批准号: 2230034
• 负责人: Wenwen Li
• 金额: $ 68.06万
• 依托单位: Arizona State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-03-01 至 2026-02-28
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2230034&HistoricalAwards=false
• 关键词: Collaborative; Research; CyberTraining; Implementation; Medium

The Arctic is one of the Earth's remaining frontiers that is critical to the Earth's climate systems. Climate change and permafrost warming are documented across the Arctic, with such warming releasing greenhouse gasses that further drive global warming. The Arctic ecosystem has been pushed to a tipping point with dramatic impacts to inland and coastal landscapes: altered soil carbon fluxes, changes in vegetation cover, erosion, shifts in animal behavior, and challenges to infrastructure. As this transformation of ice to water through degrading permafrost and melting sea and lake ice reverberates through the entire Arctic ecosystem, understanding of Arctic change necessitates research from a broad range of Earth, engineering, and social science disciplines. Valuable climatic, geological, biological, and sociological data exist but have yet to be fully exploited by the Arctic science community. Artificial Intelligence (AI) and machine learning approaches, which have the ability to automatically process big data and extract hidden knowledge, would enable researchers to make the best possible use of these data to address diverse Arctic challenges. This project will develop a novel cybertraining program to increase the capacity for myriad Arctic researchers across disciplines to employ AI-driven techniques on Arctic data. These new skills will enable current and future Arctic scientists to use the new wave of data-driven discovery tools and thereby better understand the rapidly changing Arctic landscape, which is critically needed for societal welfare.Today, Artificial Intelligence has become one of the most powerful tools to analyze big data and enable a new paradigm of data-driven science. However, training in these emerging topics is largely missing in current undergraduate and graduate curricula, as well as for active Arctic researchers. This project will foster the growth of an Arctic science workforce by developing data science skills through a series of complementary and mutually reinforcing training activities. An Arctic-AI research network will be established for collecting AI training needs and for Arctic scientists and AI experts to share ideas and resources, to network with each other, and to experience the latest research advances through a monthly webinar series. Customized training will be provided through both in-person workshops and online, self-paced learning programs to broaden the adoption of advanced AI methods in Arctic science. The workshops will be open not only to Arctic researchers, but also to the Arctic science educators, offering a pathway for interested faculty and instructors at multiple institutions to incorporate training materials into their curricula and classroom teaching, amplifying the scale of the cybertraining activities. Meanwhile, an open competition, the Arctic GeoAI Challenge, will be launched as a novel form of hands-on technology training to attract talented individuals to develop novel AI solutions for solving a real-world Arctic big data problem. The recruitment plan will cultivate an inclusive and diverse culture of community, with a strong focus on growing the STEM research workforce with more women, women of color, and people from diverse ethnic groups, academic backgrounds, and sectors, enabling especially the Arctic indigenous community to have a greater voice in understanding and mitigating Arctic change. All training materials will be deposited in the Arctic Data Center's Learning Hub and the Permafrost Discovery Gateway to ensure long-term access by cyberinfrastructure users, professionals, and developers across all Arctic science and geoscience domains and beyond. This project is co-funded by a collaboration between the Directorate for Geosciences and Office of Advanced Cyberinfrastructure to support AI/ML and open science activities in the geosciences.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

北极是地球剩余的边界之一，对地球的气候系统至关重要。气候变化和永久冻土变暖在整个北极地区都有记录，这种变暖会释放温室气体，进一步推动全球变暖。北极生态系统已被推到一个临界点，对内陆和沿海景观产生了巨大影响：土壤碳通量改变，植被覆盖变化，侵蚀，动物行为变化以及对基础设施的挑战。由于这种通过永久冻土退化和海冰和湖冰融化的冰到水的转化在整个北极生态系统中产生了反响，因此对北极变化的理解需要广泛的地球，工程和社会科学学科的研究。有价值的气候，地质，生物和社会学数据存在，但尚未被北极科学界充分利用。人工智能（AI）和机器学习方法能够自动处理大数据并提取隐藏的知识，这将使研究人员能够最好地利用这些数据来应对北极的各种挑战。该项目将开发一个新的网络培训计划，以提高众多北极研究人员在北极数据上采用人工智能驱动技术的能力。这些新技能将使当前和未来的北极科学家能够使用新一波数据驱动的发现工具，从而更好地了解快速变化的北极景观，这对社会福利至关重要。今天，人工智能已成为分析大数据的最强大工具之一，并实现数据驱动科学的新范式。然而，目前的本科生和研究生课程以及活跃的北极研究人员在很大程度上缺少这些新兴主题的培训。该项目将通过一系列互补和相互加强的培训活动，发展数据科学技能，促进北极科学劳动力的增长。将建立一个北极人工智能研究网络，收集人工智能培训需求，让北极科学家和人工智能专家分享想法和资源，相互交流，并通过每月的网络研讨会系列体验最新的研究进展。定制培训将通过现场研讨会和在线自定进度的学习计划提供，以扩大北极科学中先进人工智能方法的采用。这些讲习班不仅对北极研究人员开放，也对北极科学教育工作者开放，为多个机构感兴趣的教师和教员提供一条途径，将培训材料纳入其课程和课堂教学，扩大网络培训活动的规模。与此同时，一项名为Arctic GeoAI Challenge的公开竞赛将作为一种新的实践技术培训形式推出，以吸引有才华的个人开发新的人工智能解决方案，以解决现实世界中的北极大数据问题。招聘计划将培养一种包容性和多元化的社区文化，重点是增加STEM研究人员，增加更多女性，有色人种女性以及来自不同种族，学术背景和部门的人，特别是北极土著社区在理解和缓解北极变化方面有更大的发言权。所有培训材料将存放在北极数据中心的学习中心和永久冻土发现网关，以确保网络基础设施用户，专业人员和开发人员在所有北极科学和地球科学领域及其他领域的长期访问。该项目由地球科学理事会和高级网络基础设施办公室共同资助，旨在支持地球科学领域的人工智能/机器学习和开放科学活动。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Segment Anything Model Can Not Segment Anything: Assessing AI Foundation Model’s Generalizability in Permafrost Mapping

分割任何东西模型无法分割任何东西：评估 AI 基础模型在永久冻土绘图中的通用性

• DOI: 10.3390/rs16050797
• 发表时间: 2024
• 期刊: Remote Sensing
• 影响因子: 5
• 作者: Li, Wenwen;Hsu, Chia-Yu;Wang, Sizhe;Yang, Yezhou;Lee, Hyunho;Liljedahl, Anna;Witharana, Chandi;Yang, Yili;Rogers, Brendan M.;Arundel, Samantha T.
• 通讯作者: Arundel, Samantha T.