Webinar Series by ISPRS TC IV
Network SpaceTime AI: Concepts and Applications (approximately 1.5 hours)
Date/Time: Thursday, February 3, 11am CET (UTC +1)
ISPRS WG IV/8 (GeoComputation and GeoSimulation) Webinar
Moderator: Dr James Haworth, Co-Chair of ISPRS WG IV/8
SpaceTimeLab, Department of Civil, Environmental and Geomatic Engineering, University College London, UK
The topic of GeoAI has risen to prominence in recent years as a field of study that applies the latest methods from computer science, such as deep learning, to geospatial problems. While GeoAI methods have had great success in image processing tasks due to their natural applicability to raster data, their application to other spatial and spatiotemporal data types remains underexplored.
Part 1 - Prof Tao Cheng
SpaceTimeLab, Department of Civil, Environmental and Geomatic Engineering, University College London, UK
In the first part of this seminar, Prof Tao Cheng will describe how networks (graphs) can be used as the spatial structure to present space-time processes happening at places that are conventionally represented as spatial units: points, lines/networks, or polygon/areas/grids. These processes could be dense (everywhere) or sparse (only somewhere), such as flows on all roads or crimes at irregular locations. The talk will showcase applications in transport, crime and public health to demonstrate the strength of using networks and graph-based deep learning for space-time prediction and analysis.
Part 2 - Seula Park
Department of Civil and Environmental Engineering, Seoul National University, South Korea
The second part of the seminar will deal with practical issues of working with large network datasets. Seula Park will introduce the graph database Neo4j as a way of representing spatial networks, before showing some real-world case studies in applications such as indoor navigation. The talk aims to highlight both the advantages of graph databases over relational databases as well as the areas where further research is needed.