The article reviews significant advances in networked signal and information processing, which have enabled in the last 25 years extending decision making and inference, optimization, control, and learning to the increasingly ubiquitous environments of distributed agents. As these interacting agents cooperate, new collective behaviors emerge from local decisions and actions. Moreover, and significantly, theory and applications show that networked agents, through cooperation and sharing, are able to match the performance of cloud or federated solutions, while offering the potential for improved privacy, increasing resilience, and saving resources.
翻译:本文回顾了网络信号和信息处理的重大进展,这些进展使得在过去25年中,决策和推断、优化、控制和学习扩展到了分布式Agent的越来越普遍的环境中。随着这些互动代理人的合作,将从本地的决策和行动中出现新的集体行为。此外,理论和应用表明,通过合作和分享,网络代理能够匹配云或联合解决方案的性能,同时提供提高隐私、提高韧性和节省资源的潜力。