CAREER: Real-Time Sampling, Estimation, and Inference in Networked Systems

物联网和社交网络为实时信息的生成、传播和收集提供了前所未有的平台。信息通常由随时间和/或空间（例如，在底层网络上）发展的过程来管理。这些过程的估计和推理，特别是在时间关键应用中，需要有效的实时和自适应采样策略。该项目将为网络系统中的实时采样、估计和推理提供理论基础和算法设计，并考虑物联网中的估计和控制、疾病（如COVID-19）的检测和检疫、及时发现和控制社交网络中错误信息的传播等应用。该项目是跨学科的，具有协同计划，以丰富宾夕法尼亚大学的研究生和本科课程，目标是缩小理论与实践的差距。它将进一步为本科生提供研究机会，特别是针对代表性不足的少数民族。外展计划包括通过美国国家科学基金会资助的创新队项目在学术界和产业界之间建立伙伴关系，通过美国国家科学基金会资助的向上数学科学项目改善STEM教育和对低收入和第一代学生的指导，公众素养和参与有关错误信息的传播。该项目的总体目标是开发实时（顺序）采样策略的基础框架，利用网络中的部分信息进行决策。它建立了信息提取和奖励最大化之间的基本权衡，前者传统上根植于信息论，后者根植于网络，并通过融合信息论、随机过程、网络科学、图形模型和机器学习的思想来开发解决方案。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 Internet of Things (IoT) and social networks have provided unprecedented platforms for the generation, dissemination and collection of real-time information. The information is often governed by processes that evolve over time and/or space (for example, on an underlying network). Estimation and inference for such processes, especially in time-critical applications, require efficient real-time and adaptive sampling strategies. This project will develop theoretical foundations and algorithmic designs for real-time sampling, estimation and inference in networked systems and considers applications such as estimation and control in IoT, testing and quarantine for diseases (e.g. COVID-19), and timely detection and control of the spread of misinformation in social networks. The project is interdisciplinary and features a synergistic plan to enrich the graduate and undergraduate curriculum at the University of Pennsylvania with the goal of narrowing the theory-practice gap. It will further provide undergraduates with research opportunities, targeting in particular underrepresented minorities. The outreach plan includes partnerships between academia and industry through the NSF-funded Innovation Corps program, public literacy and engagement concerning the spread of misinformation, as well as improved STEM education and mentorship of low-income and first-generation students through the NSF-funded Upward Bound Math Science program.The overarching goal of the project is to develop foundational frameworks for real-time (sequential) sampling strategies that exploit partial information for decision making in networks. It establishes fundamental tradeoffs between information extraction and reward maximization with the former traditionally rooted in information theory and the latter in networking, and develops solutions by merging ideas from information theory, stochastic processes, network sciences, graphical models, and machine learning.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.