CIF: Large: Collaborative Research: Cooperation and Learning over Cognitive Networks

认知网络上的合作和学习经济学和社会科学和生物科学中关于羊群和自组织的研究已经观察到，多个主体之间的协调会导致规律的行为模式和群体智能，即使每个群体成员表现出有限的行为复杂性。例如，在蚂蚁群体中，由于其有限的感知能力，单个蚂蚁无法从其环境中捕获丰富的空间信息。然而，当蚂蚁在一个群体内协调它们的活动时，这个群体最终表现出更好的感知能力。利用信号处理和通信技术，研究如何以及为什么这种理性和有组织的行为在群体层面上由能力有限的代理之间的局部交互产生，什么通信拓扑实现了这种行为，以及什么类型的信号处理实现了这种形成。这项研究试图通过调查认知网络与学习和理性的关系，来理解和逆向工程社会经济生物网络中遇到的分布式智能。后者是自适应网络，避免了集中的信息处理，并执行网内推理和控制决策。认知网络与依赖于集中式和并行信息融合的网络相比，具有不可伸缩性，难以适应不断变化的拓扑结构，存在脆弱性和信息瓶颈。这项研究考虑了大规模的代理人网络，并研究了全球(理性或非理性)行为模式是如何出现的，包括经济中的羊群、传染病和泡沫。对生物环境如何影响动物社会中的集体行为的理解，为良好的认知网络提供了现实世界的指南，这些网络可以反过来用于设计工程系统。认知网络在精准农业、环境监测、救灾管理和智能空间等领域都有应用。

Cooperation and Learning over Cognitive NetworksStudies on herding and self-organization in economics and the social and biological sciences have observed that coordination among multiple agents leads to regular patterns of behavior and swarm intelligence, even when each group member shows limited behavioral complexity. In ant colonies, for example, individual ants cannot capture rich spatial information from their environment because of their limited sensing ability. Nevertheless, when the ants coordinate their activities within a colony, the group ends up exhibiting better sensing abilities. Using signal processing and communications techniques, the research studies how and why such manifestations of rational and organized behavior arise at the group level from local interactions among agents with limited abilities, what communication topologies enable such behavior, and what type of signal processing enables such formations. This research seeks to understand and reverse-engineer the distributed intelligence encountered in socio-economic-biological networks, by investigating relations with learning and rationality over cognitive networks. The latter are adaptive networks that avoid centralized information processing and perform in-network inference and control decisions. Cognitive networks contrast with networks that rely on centralized and parallel information fusion, which are not scalable, are hard to adapt to changing topologies, and suffer from points of vulnerability and information bottlenecks. The research considers large scale networks of agents and studies how global (rational or irrational) patterns of behavior emerge, including herds, contagions and bubbles in economics. An understanding of how the biotic environment influences collective behavior in animal societies provides a real world guide to good cognitive networks, which can be used in turn to design engineered systems. Cognitive networks have applications in areas ranging from precision agriculture, to environmental monitoring, disaster relief management, and smart spaces.