The Role of Agency and Interactivity in Embodied Conversational Agents to Support Teacher Training in High Poverty Schools

代理和交互性在具体对话代理中的作用，以支持高贫困学校的教师培训

• 批准号: 2119265
• 负责人: Elisa Shernoff
• 金额: $ 84.99万
• 依托单位: Rutgers University New Brunswick
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-09-01 至 2024-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2119265&HistoricalAwards=false
• 关键词: Role; Agency; Interactivity; Embodied; Conversational

Challenges with classroom behavior management predicts teacher stress and attrition and support for behavior management is a top professional priority identified by teachers. Traditional training for teachers in behavior management (e.g., workshops, whole group in-service training) has failed to produce substantial improvements in teachers’ work-related knowledge and skills. Virtual Reality-based training provides an opportunity for teachers to practice responding positively to disruptive behaviors with the freedom to explore and fail in a low-stakes setting. However, existing platforms suffer from low usability and limited realism, particularly when simulating conversational interactions. This project offers a synergistic exploration of the educational impact of immersive, conversational learning environments, and the fundamental technical capabilities for building intelligent virtual agents. Specifically, the aim of the project is to systematically study the impact of embodied conversational agents on learning outcomes of teachers in high poverty schools, where discipline disparities and exclusionary discipline practices are prominent and positive discipline approaches are needed to promote engagement and learning.This project supports the development and release of a new platform for specifying embodied conversational interaction, combining recent advances in intelligent virtual agents, generative modeling of communicative behavior, and human-in-the-loop design. The project outcomes will push the boundaries of agency and interactivity in virtual reality training platforms, where teachers can fully immerse themselves in virtual classrooms and engage with autonomous virtual students capable of expressing themselves using facial expressions, hand and body gestures, and spoken language. These free-form interactions are theorized to support transfer of learning from one setting (e.g., virtual classroom) to a new setting (e.g., live classroom), where teachers may carry forward applicable knowledge and skills. The research plan includes a series of studies over a 3-year time period, to systematically study the impact of agency and interactivity in virtual training platforms on the training outcomes of teachers, and their ability to transfer knowledge and skills to real classrooms. This type of training in behavior management provides an unique opportunity to augment existing support to teachers and foster the dissemination and implementation of effective training that can ultimately enhance student achievement and student/teacher well-being.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.

课堂行为管理的挑战预示着教师的压力和流失，对行为管理的支持是教师确定的最高专业优先事项。传统的教师行为管理培训（例如，工作坊、整群在职培训等）并没有实质性地提高教师的工作知识和技能。基于虚拟现实的培训为教师提供了一个机会，让他们练习积极应对破坏性行为，并在低风险环境中自由探索和失败。然而，现有的平台遭受低可用性和有限的现实主义，特别是当模拟会话交互。该项目提供了沉浸式，对话式学习环境的教育影响的协同探索，以及构建智能虚拟代理的基本技术能力。具体而言，该项目的目的是系统地研究体现性对话代理对高度贫困学校教师学习成果的影响，这些学校的学科差异和排斥性学科实践突出，需要积极的学科方法来促进参与和学习。该项目支持开发和发布一个新的平台，用于指定体现性对话互动，结合了智能虚拟代理，通信行为的生成建模和人在回路设计的最新进展。该项目的成果将推动虚拟现实培训平台中的代理和交互边界，教师可以完全沉浸在虚拟教室中，并与能够使用面部表情，手势和身体姿势以及口语表达自己的自主虚拟学生进行互动。这些自由形式的交互被理论化以支持从一个环境（例如，虚拟教室）到新的设置（例如，现场教室），教师可以在那里传授适用的知识和技能。该研究计划包括一系列为期3年的研究，以系统地研究虚拟培训平台中的代理和互动对教师培训成果的影响，以及他们将知识和技能转移到真实的课堂的能力。这种行为管理培训提供了一个独特的机会，以增加现有的支持教师和促进传播和实施有效的培训，最终可以提高学生的成绩和学生/教师wellbeing.This奖项反映了NSF的法定使命，并已被认为是值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估的支持。

项目成果

COMPOSER: Compositional Reasoning of Group Activity in Videos with Keypoint-Only Modality

• DOI: 10.1007/978-3-031-19833-5_15
• 发表时间: 2021-12
• 作者: Honglu Zhou;Asim Kadav;Aviv Shamsian;Shijie Geng;Farley Lai;Long Zhao;Tingxi Liu;M. Kapadia

Harnessing Fourier Isovists and Geodesic Interaction for Long-Term Crowd Flow Prediction

利用傅里叶等量线和测地线相互作用进行长期人群流量预测

• DOI: 10.24963/ijcai.2022/185
• 发表时间: 2022
• 期刊: Thirty-First International Joint Conference on Artificial Intelligence (IJCAI
• 作者: Sohn, Samuel S.;Moon, Seonghyeon;Zhou, Honglu;Lee, Mihee;Yoon, Sejong;Pavlovic, Vladimir;Kapadia, Mubbasir
• 通讯作者: Kapadia, Mubbasir

The Importance of Multimodal Emotion Conditioning and Affect Consistency for Embodied Conversational Agents

• DOI: 10.1145/3581641.3584045
• 发表时间: 2023-03
• 期刊: Proceedings of the 28th International Conference on Intelligent User Interfaces
• 作者: Che-Jui Chang;Samuel S. Sohn;Sen Zhang;R. Jayashankar;Muhammad Usman;M. Kapadia

An Integrated Architecture for Common Ground in Collaboration

协作共同基础的集成架构

• 发表时间: 2022
• 期刊: http://www.cogsys.org/
• 作者: Geib, Christopher;George, Denson;Khalid, Baber;Magnotti, Richard;Stone, Matthew
• 通讯作者: Stone, Matthew

HM: Hybrid Masking for Few-Shot Segmentation

• DOI: 10.1007/978-3-031-20044-1_29
• 发表时间: 2022-03
• 作者: Seonghyeon Moon;Samuel S. Sohn;Honglu Zhou;Sejong Yoon;V. Pavlovic;Muhammad Haris Khan;M. Kapadia