GeoAI and Deep Learning Symposium: Deep Learning of Geospatial Patterns & Applications

Type: Paper
Theme:
Sponsor Groups: Spatial Analysis and Modeling Specialty Group, Geographic Information Science and Systems Specialty Group, Cyberinfrastructure Specialty Group
Poster #:
Day: 4/5/2019
Start / End Time: 5:00 PM / 6:40 PM
Room: Washington 6, Marriott, Exhibition Level
Organizers: Rui Zhu, Kevin Mwenda, Guofeng Cao
Chairs: Rui Zhu

Description

In the past decade, deep learning has become one of the most successful techniques in studying patterns thanks to the increasing power of modern computations. Multiple scientific domains such as computer vision, text mining and speech recognition significantly benefit from the advent of deep learning. Geographers also start to embrace the deep learning in analyzing geospatial patterns, example geospatial applications include object detection in remotely sensed images, trajectory and network analysis, geospatial semantic analysis, and so on. However, the nature of geospatial patterns is distinguishable from general patterns. Therefore, specifications and/or adaptations on applying deep learning are necessary to tackle geospatial problems. In addition, geographers also endeavor to contribute to deep learning by introducing techniques from geographic information sciences such as geostatistics, spatial optimization and spatial cognition. Specifically, researchers are interested in answering fundamental questions like: How can deep learning be employed and adapted to analyze and predict geographic phenomena (e.g. urban expansion & evolution)? What are the conceptual and fundamental considerations behind building and training neural networks to conduct geospatial data analysis? What are the mechanics underlying deep learning methods and how can these techniques be leveraged by geographers to uncover patterns and structure embedded in geographic data? How can theories and/or techniques (e.g., spatial statistics, spatial optimization) developed in GIScience be adapted to improve the performance of deep learning?

To advance this research direction, we invite researchers to present their work that are relevant to topics that include, but are not limited to:
(1) Methodologies to improve the performance of deep learning of geospatial patterns
(2) Sensitivity analysis of deep learning on understanding geospatial patterns
(3) Applications of deep learning from remotely sensed images
(4) Applications of deep learning of geospatial networks/trajectories
(6) Using deep learning to analyze and predict urban land use/land cover (LULC) changes


Agenda

Type Details Minutes Start Time
Presenter Bo Yan*, University of California, Santa Barbara, Krzysztof Janowicz, University of California, Santa Barbara, Gengchen Mai, University of California, Santa Barbara, Rui Zhu, University of California, Santa Barbara, xNet+SC: Classifying Places Based on Images by Incorporating Spatial Contexts 20 5:00 PM
Presenter Di Zhu*, Peking University, Ximeng Cheng, Peking University, Fan Zhang, Massachusetts Institute of Technology, Yong Gao, Peking University, Yu Liu, Peking University, Spatial interpolation based on conditional generative adversarial neural network 20 5:20 PM
Presenter Mohammad Eshghi*, University of Oregon, Machine Learning Predictive Modeling for Geospatial Occurrence-only Data 20 5:40 PM
Presenter Firoozeh Karimi*, University of north carolina at greensboro, Selima Sultana, University of north carolina at greensboro, Exploration of Decision Tree Algorithm for Modeling Urban Expansion patterns 20 6:00 PM
Presenter Karim Malik*, , Colin Robertson, Wilfrid Laurier University, Landscape pattern similarity comparison: the role of texture-encoded convolutional neural networks 20 6:20 PM

To access contact information login