Academic Report on July.31,2019(2)

Reporter Topic:Bayesian supervised learning of spatially varying but sparsely measured geo-data

Reporter: Professor Yu Wang, City University of Hong Kong

Time:14:00, July.31, 2019

Location:Conference Room 8319, Department of Engineering

    Spatial data (i.e., data depending on spatial coordinates, such as geographic locations) are ingredients of many important applications, including building information modeling (BIM) and smart city. Examples of spatial data include geotechnical data, seismic data, wind data, traffic data, hydrologic data, geological data, air quality data, soil   & water contamination data. Although spatial data are spatially varying and correlated, they are often sparsely measured due to time, resource, or technical constraints. The seminar presents some emerging methods for effective interpretation of sparsely measured spatial data using Bayesian supervised learning (BSL) and compressive sensing (CS). The BSL and CS results can also be used together with Karhunen–Loève (KL) expansion for random field modeling. Some insights into the random field modeling of site-specific spatial variability will be discussed.