Excellence in Research:Towards Data and Machine Learning Fairness in Smart Mobility

卓越研究：实现智能移动中的数据和机器学习公平

• 批准号: 2401655
• 负责人: Di Yang
• 金额: $ 59.2万
• 依托单位: Morgan State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2024
• 资助国家: 美国
• 起止时间: 2024-08-01 至 2027-07-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2401655&HistoricalAwards=false
• 关键词: Excellence; Research; Towards; Data; Machine

This project supports research examining the development of fairness-aware methodologies to address prevalent data and machine learning (ML) biases within smart mobility applications. With the advancements in intelligent sensors and computing power, the integration of high-fidelity transportation data with Artificial Intelligence (AI)/ML has become essential for advancing smart mobility applications. This project aims to investigate ways to promote fair, equitable, and responsible AI utilization in tackling diverse smart mobility challenges, such as vehicle trajectory prediction, congestion reduction, safety improvement, and so on. With the primary institution being Morgan State University, an R2 public Historically Black College and University (HBCU), this project fosters research engagement among undergraduate and graduate students, with a focus on individuals from historically marginalized backgrounds. Furthermore, to prepare the future workforce for the evolving technological landscapes in transportation, this project serves as a bridge by connecting STEM learning from K-12 through post-secondary education with cutting-edge AI/ML methods and their applications in smart mobility.This project aims to investigate development of fairness-aware methodologies to mitigate commonly encountered data and ML biases that are often induced from data collection, processing, and modeling within the smart mobility domain. Specifically, this project targets three critical biases throughout the ML application lifecycle: measurement bias, representation bias, and aggregation bias. Customized ML methodologies are devised to mitigate each type of biases, tailored for specific smart mobility applications, including vehicle trajectory correction and prediction, traffic flow and network modeling, origin-destination and traffic demand forecasting, among others. Potential findings from this project can promote fair and equitable applications of ML methods in smart mobility and can have broad impacts on other science and engineering fields, such as smart and autonomous systems, robotics, and other research domains that depend on the responsible utilization of AI/ML. Students from underrepresented groups, particularly African-American students at Morgan State University, are strongly encouraged to participate in the research.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.

该项目支持研究开发公平意识方法，以解决智能移动应用中普遍存在的数据和机器学习 (ML) 偏差。随着智能传感器和计算能力的进步，高保真交通数据与人工智能 (AI)/ML 的集成已成为推进智能移动应用的关键。该项目旨在研究如何促进公平、公正和负责任的人工智能应用，以应对各种智能移动挑战，例如车辆轨迹预测、减少拥堵、提高安全性等。该项目的主要机构是摩根州立大学，这是一所 R2 公立历史黑人学院和大学 (HBCU)，旨在促进本科生和研究生的研究参与，重点关注来自历史边缘背景的个人。此外，为了让未来的劳动力为不断发展的交通技术格局做好准备，该项目充当了一座桥梁，将 K-12 到高等教育的 STEM 学习与尖端的 AI/ML 方法及其在智能移动中的应用连接起来。该项目旨在研究公平意识方法的开发，以减轻智能移动领域内的数据收集、处理和建模中经常遇到的常见数据和机器学习偏差。具体来说，该项目针对整个 ML 应用程序生命周期中的三个关键偏差：测量偏差、表示偏差和聚合偏差。定制的机器学习方法旨在减轻每种类型的偏差，针对特定的智能移动应用量身定制，包括车辆轨迹校正和预测、交通流和网络建模、起点-目的地和交通需求预测等。该项目的潜在发现可以促进机器学习方法在智能移动领域的公平和公正应用，并对其他科学和工程领域产生广泛影响，例如智能和自主系统、机器人技术以及其他依赖于负责任地利用人工智能的研究领域/ML。强烈鼓励来自代表性不足群体的学生，特别是摩根州立大学的非裔美国学生参与这项研究。该奖项反映了 NSF 的法定使命，并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，被认为值得支持。