CyberTraining: Pilot: Machine Learning Foundations and Applications in the Earth Systems Sciences

在地球系统科学(ESS)中越来越多地使用机器学习技术对科学和工程研究产生了积极影响，但易用性加上许多常用工具的不透明性质，可能导致在不对工具的输出进行严格评估的情况下产生信任。围绕机器学习工具的理论基础的教育需要额外的、往往是实质性的高等数学和编程课程。无法获得这一背景信息的网络基础设施用户可能难以欣赏和理解机器学习技术对其研究和职业目标的实用性和适当性，特别是当他们就读于资源不足的机构时，这些机构无法自行创建相关的教育材料。该项目旨在帮助大学水平的学习者建立必要的网络基础设施、知识和技能，使他们能够适当地将机器学习技术应用到他们的ESS研究中，而不需要进行令人望而却步的额外课程。这些未来的网络基础设施用户受益于对机器学习技术的适当和合乎道德的使用，即使正在使用的工具是由其他人开发的。该项目的目标是为大学水平的ESS学生和职业生涯早期的专业人员阐明机器学习模型背后的概念机制，并弥合机器学习概念与地球系统科学中低代码、真实世界应用之间的差距。它将通过提供一系列三个学习模块来实现这一点：(1)自定进度的概念介绍，使用系统思维方法来理解机器学习在ESS中的工作方式；(2)自定进度的低代码模块，使学习者能够将概念框架应用于具有相关ESS数据的真实世界场景；以及(3)基于实验室的活动，促进小组讨论、决策制定的合理性以及对机器学习技术和输出的批判性分析。这种设计很适合与现有课程整合的翻转课堂环境，允许学习者练习这些技能，而不需要承担额外的课程作业。此外，该计划培养ESS网络基础设施用户的关键判断技能，促进网络基础设施和机器学习的广泛、战略、道德和适当使用。该奖项由高级网络基础设施办公室联合地球科学局支持。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The increasing use of machine learning techniques in Earth Systems Science (ESS) has positive impacts on science and engineering research, but ease of use coupled with the opaque nature of many commonly available tools can lead to trust without critical assessment of the tools' outputs. Education surrounding the theoretical underpinnings of machine learning tools requires additional, and often substantive, coursework in advanced math and programming. Cyberinfrastructure users without access to this background information may have difficulty building an appreciation and understanding of the utility and appropriateness of machine learning technology for their research and career goals, especially when they attend under-resourced institutions that cannot create relevant educational materials on their own. This project intends to help university-level learners build necessary cyberinfrastructure literacy and skills that will allow them to appropriately apply machine learning techniques to their ESS research without requiring prohibitive additional coursework. These future cyberinfrastructure users benefit from practicing appropriate and ethical usage of machine learning techniques even when the tools in use were developed by others. The goals of this project are to elucidate the conceptual mechanisms behind machine learning models for university-level ESS students and early-career professionals, and to bridge the gap between machine learning concepts and low-code, real-world applications in the Earth Systems Sciences. It will accomplish this by providing a series of three learning modules: (1) a self-paced conceptual introduction that uses a systems-thinking approach to understanding how machine learning works in ESS, (2) a self-paced, low-code module that enables learners to apply the conceptual frameworks to real world scenarios with relevant ESS data, and (3) a lab-based activity that promotes group discussion, justification of decision making, and critical analysis of machine learning techniques and outputs. This design lends itself well to a flipped classroom setting integrated with existing curricula, allowing learners to practice these skills without the need to take on additional coursework. Additionally, the program fosters critical judgment skills in ESS cyberinfrastructure users, furthering the broad, strategic, ethical, and appropriate usage of cyberinfrastructure and machine learning.This award by the Office of Advanced Cyberinfrastructure is jointly supported by the Directorate for 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.