Predicting Human Contacts Through Alternating Direction Method Of Multipliers
INTERNATIONAL JOURNAL OF MODERN PHYSICS C(2019)
摘要
Transmission of respiratory infectious diseases depends greatly on human close-proximity contacts, making thorough understanding of current and upcoming contacts essential for epidemic containment. Although different devices and software have been developed for contact data collection, there are few effective methods for contact prediction available in the near future as far as the authors know. In this study, we propose an approach to predict human contacts. We first extract human features together with their significances from the human contacts through alternating direction method of multipliers (ADMM), then predict future significances based on periodicity of contacts, and finally construct future contacts from human features and future significances. With the help of contact data collected in a Chinese University, we compare this approach with a trivial method of directly averaging known contacts. The comparison shows that our approach generates contacts deviating less from the true ones.
更多查看译文
关键词
Human contact prediction, alternating direction method of multipliers, periodicity
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络