Convergence: RAISE: A Flexible Framework for Instrumented Learning Environments: Enhanced Learning Through Advanced Sensing, Processing, and Cognitive Technologies

美国国家工程院将推进个性化学习确定为其14大工程挑战之一。为了在这一挑战上取得进展，一个拥有认知和学习科学、物理教育、计算机系统、生物传感和机器学习专业知识的研究团队将共同努力，带来先进的数据采集和处理技术，以支持对认知和学习科学假设的快速测试和评估。该项目将采用一种融合的方法来推进主动学习的研究，以及在物理入门课程中使用主动学习的课堂压力的作用。这项努力集中在两个目标上。首先是开发技术，建立一个灵活的框架，用于创建用于大学物理入门的测量压力的仪器。第二项是开展试点研究，以建立一种基于认知和学习科学的联合理解，了解压力的作用和学习机制在学生多视角对话（MPCs）背景下的学习。虽然mpc在发展物理学等学科的概念理解方面发挥着重要作用，但人们对mpc期间刺激和压力的影响知之甚少。因为几十个mpc同时发生在教室里，所以记录和研究它们是具有挑战性的。该项目开发的自动记录和机器学习工具将允许收集和研究更多的对话。使用先进生物传感器的认知科学实验室实验将首先在受控环境中研究压力和学习。随后，生物传感器和音频/视频数据的混合将使压力和学习问题首次在更大的课堂环境中得到研究。该项目的一个重要成果将是建立一个跨学科的融合团队。活动将集中在建立共享词汇、跨学科会议、公共讲习班和通过研究生课程和参与该项目培训未来的融合研究人员。该奖项由OIA、EHR、ENG和SBE资助。技术开发工作分为两个方面，每个方面都有一个由认知科学家、学习科学家和工程师组成的跨学科团队。Aspect I研究团队将开发先进的生物传感器，通过不显眼的可穿戴贴片和教室皮质醇传感器来测量压力。这些传感器将用于认知科学实验室，通过测量注意力、即时记忆和构建知识图式的能力来研究压力对学习的影响。在第二方面，将开发基于拓扑数据分析的机器学习方法，将音频/视频记录和生物传感器数据结合起来，使学习和认知科学家能够大规模地研究主动学习环境。这些试点工作的重点将放在物理导论课堂的多视角对话（MPCs）研究上。用于测量学习环境的灵活框架，包括压力传感器、自动转录和机器学习工具，将首先在少量并发mpc上进行测试，然后在教室规模上进行测试。研究人员将验证该技术，并开始调查MPC与学习成果之间的关系，培养学生之间MPC的条件，以及压力在形成MPC基础的认知过程中的作用。这种基于线的传感器将使用新型材料，并且不显眼、可穿戴和无线。拓扑数据分析（TDA）的使用将提供一个灵活的机器学习框架，该框架可以随着数据集的大小和异质性而增长和适应。这些技术将被计算机系统研究人员整合并部署到一个教室规模的系统中。该奖项反映了美国国家科学基金会的法定使命，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。

The National Academy of Engineering identified advancing personalized learning as one of its 14 Grand Challenges for Engineering. To make progress on this challenge, a research team with expertise in cognitive and learning sciences, physics education, computer systems, bio-sensing, and machine learning will work together to bring advanced data acquisition and processing technologies to support rapid testing and evaluation of cognitive and learning science hypotheses. The project will use a convergent approach to advance the study of active learning and the role of stress in classes using active learning in introductory physics. The effort is centered on two objectives. The first is to develop technology to build a flexible framework for creating instruments to measure stress for use in introductory college physics. The second is to pilot studies to develop a joint cognitive and learning science-based understanding of the role of stress and the mechanisms for learning in the context of multi-perspective conversations (MPCs) among students in introductory physics. While MPCs are known to play an important role in developing conceptual understanding in disciplines such as physics, little is understood concerning the impact of stimulation and stress during MPCs. Because dozens of MPCs occur simultaneously in a classroom, they are challenging to record and study. The automated transcript and machine learning tools developed by this project will allow for the collection and study of significantly more conversations. Cognitive science lab experiments using advanced bio-sensors will study stress and learning initially in a controlled setting. Subsequent blending of biosensor and audio/video data will allow the issues of stress and learning to be studied for the first time in a larger classroom environment. One of the key broader outcomes of the project will be the construction of an interdisciplinary convergent team. Activities will focus around building shared vocabulary, cross-disciplinary meetings, public workshops and the training of future convergent researchers through graduate courses and participation in this project. The award is supported by funding from OIA, EHR, ENG, and SBE. The technology development efforts are divided into two aspects each with an interdisciplinary team of cognitive scientists, learning scientists and engineers. The Aspect I research team will develop advanced bio-sensors to measure stress through unobtrusive wearable patches and classroom cortisol sensors. These sensors will be used in the cognitive science lab to study the effects of stress in learning by measuring attention, immediate memory, and the ability to construct knowledge schemas. In Aspect II, machine learning methods based on topological data analysis will be developed that combine audio/video recording and the biosensors data to allow learning and cognitive scientists to study active learning environments at scale. The focus of these pilot efforts will be on the study of multi-perspective conversations (MPCs) in an introductory physics classroom. Flexible framework for instrumenting learning environments, complete with stress sensors and automated transcripts and machine learning tools, will be tested first with a small number of concurrent MPCs and then later at classroom scale. Researchers will verify the technology and begin investigating the relationship between MPC and learning outcomes, the conditions that foster MPCs between students, and the role of stress in shaping cognitive processes underlying MPCs. The thread-based sensors will use novel materials and are non-obtrusive and wearable and wireless. The use of topological data analysis (TDA) will provide a flexible machine learning framework that can grow and adapt both with the size of the data sets and their heterogeneity. These technologies will be integrated and deployed in a classroom-scale system by computer systems researchers.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.