Aims and Scope
The significant advancements in software and hardware technologies stimulated the prosperities of the domains in spatial computing and deep learning algorithms, respectively. Recent breakthroughs in the deep learning field have exhibited outstanding performance in handling data in space and time in specific domains such as image, audio, and video. Meanwhile, the development of sensing and data collection techniques in relevant domains have enabled and accumulated large scale of spatiotemporal data over the years, which in turn has led to unprecedented opportunities and prerequisites for the discovery of macro- and micro- spatiotemporal phenomena accurately and precisely. The complementary strengths and challenges between spatiotemporal data computing and deep learning in recent years suggest urgent needs to bring together the experts in these two domains in prestigious venues, which is still missing until now.
This workshop will provide a premium platform for both research and industry to exchange ideas on opportunities, challenges, and cutting-edge techniques of deep learning in spatiotemporal data, applications, and systems.
Topics of Interest: We encourage submissions of papers that fall into (but not limited to) the following three broad categories:
Novel Deep Learning Techniques for Spatial and Spatio-Temporal Data:
Spatial representation learning and deep neural networks for spatio-temporal data and geometric data
Physics-guided and interpretable deep learning for spatial-temporal dataDeep generative models for spatio-temporal data
Deep reinforcement learning for spatio-temporal decision making problems
Novel Applications of Deep Learning Techniques to Spatio-temporal Computing Problems. :
Remote sensing imagery and point cloud analysis in Earth science (e.g., hydrology, agriculture, ecology, natural disasters, etc.)
Deep learning for mobility and traffic data analyticsLocation-based social network data analytics, geosocial media data mining, spatial event prediction and forecasting, geographic knowledge graphs
Learning for biological data with spatial structures (bio-molecule, brain networks, etc.)
Challenges, Opportunities, and Early Progress in Deep Learning for COVID-19
Novel Deep Learning Systems for Spatio-temporal Applications:
Real-time decision-making systems for traffic management, crime prediction, accident risk analysis, etc.
GIS systems using deep learning (e.g., mapping, routing, or Smart city)Mobile computing systems using deep learningGeoAI Cyberinfrastructure for Earth science applicationsInterpretable deep learning systems for spatio-temporal temporal data
McKnight Distinguished University Professor
Vice President and Chief Scientist at Beike
Yuanqi Du, George Mason University
Arnold Boedihardjo, DigitalGlobe
Wei Wang, Microsoft Research
Ray Dos Santos, Army Corps of Engineers
Chao Zhang, Georgia Tech
Yanjie Fu, UCF
Xuchao Zhang, NEC Lab North America
Yanfang Ye, Case Western Reserve University
Yanhua Li, WPI
Jing Dai, Google
Yiqun Xie, University of Maryland
Junbo Zhang, JD Digital
Jie Bao, JD Digital
Song Gao, University of Wisconsin, Madison
Jingyuan Wang, Beihang University
Lexie Yang, ORNL
Borko Furht, Florida Atlantic University
Taghi Khoshgoftaar, Florida Atlantic University
Important Dates: (all due Midnight Anywhere on Earth).
Paper Submission: May 20, 2021
Paper Review Begins: May 21, 2021
Paper Review Due: June 5, 2021
Notification of Acceptance: June 10, 2021
Camera-ready Papers: June 17, 2021
Workshop Date: TBD
The workshop welcomes the two types of submissions
Full research papers – up to 9 pages (8 pages at most for the main body and the last page can only hold references)
Vision papers and short system papers - up to 5 pages (4 pages at most for the main body and the last page can only hold references)
All manuscripts should be submitted in a single PDF file including all content, figures, tables, and references, following the format of KDD conference papers. Paper submissions need to include author information (review not double blinded).
Papers should be submitted at: https://easychair.org/conferences/?conf=deepspatial21
Concurrent submissions to other journals and conferences are acceptable. Accepted papers will be presented as posters or short talks during the workshop and published on the workshop website. Besides, a small number of accepted papers may be selected to be presented as contributed talks. As a tradition, accepted workshop papers are NOT included in the ACM Digital Library. The authors maintain the copyright of their papers.
Rahul Ghosh, Xiaowei Jia, Chenxi Lin, Zhenong Jin, Vipin Kumar.
Shuhui Gong, Xiaopeng Mo, Rui Cao, Yu Lin, Wei Tu, Ruibin Bai
Tahereh Arabghalizi, Xiaowei Jia
Silvia Casacuberta, Esra Suel, Seth Flaxman
Tianyuan Huang, Zhecheng Wang, Hao Sheng, Andrew Y. NG, Ram Rajagopal
Yan Li, Majid Farhadloo, Santhoshi Krishnan, Timothy L Frankel, Shashi Shekhar, Arvind Rao
All the following times are in Singapore Standard Time
|9:00am-10:00am||Keynote Speech #1: What is special about Spatial Data Science and GeoAI (by Shashi Shekhar)
|10:10am-11:10am||Keynote Speech #2: Graph Convolutional Neural Networks for Physics-informed AI Models (by Yan Liu)
|11:55am-12:00pm||Awards and Closing Remarks|
The importance of spatial data science and Geo-AI is growing with the rise of spatial and spatiotemporal big data (e.g., trajectories, remote-sensing images, census and geo-social media). Societal use cases include Agriculture ( global crop monitoring, precision agriculture), Location-based services (e.g., navigation, ride-sharing), Public Health (e.g., monitoring disease spread), Environment and Climate (change detection, land-cover classification), Smart Cities (e.g., mapping buildings), etc. Classical data science and AI (e.g., machine learning) often perform poorly when applied to spatial data sets because of the many reasons. First, spatial data is embedded in a continuous space and classical statistics (e.g., correlation) are not robust to the modifiable areal unit problem. Second, spatial data-items have extended footprints (e.g., line strings, polygons) and implicit relationships (e.g., distance, touch). Third, high cost of spurious patterns requires guardrails (e.g., statistical significance tests) to reduce false positives. Furthermore, spatial autocorrelation and variability violate the classical assumption of data samples being generated independently from identical distributions, which risk models that are either inaccurate or inconsistent with the data. Thus, new methods are needed to analyze spatial data. This talk surveys common and emerging methods for spatial classification and prediction (e.g., spatial autoregression, spatial variability aware neural networks), as well as techniques for discovering interesting, useful and non-trivial patterns such as hotspots (e.g., circular, linear, arbitrary shapes ), interactions (e.g., co-locations , cascade , tele-connections ), spatial outliers, and their spatio-temporal counterparts.
Dr. Shashi Shekhar, a McKnight Distinguished University Professor at the University of Minnesota and an U.C. Berkeley alumnus, is a leading scholar of spatial computing and Geographic Information Systems (GIS). He is serving on the Computing Research Association (CRA) board, and as a co-Editor-in-Chief of Geo-Informatica journal (Springer). Earlier, he served as the President of the University Consortium for GIS (UCGIS), and on many National Academies' committees. Recognitions include IEEE-CS Technical Achievement Award, UCGIS Education Award, IEEE Fellow and AAAS Fellow. Contributions include algorithms for evacuation route planning and spatial pattern (e.g., colocation, linear hotspots) mining, an Encyclopedia of GIS, a Spatial Databases textbook, and a spatial computing book for professionals.
Deep learning has achieved significant successes in prediction performance by learning latent representations from data-rich applications, but we are confronted with many challenging learning scenarios in modeling natural phenomena, where a limited number of labeled examples are available or there is much noise in the data. Furthermore, there could be constant changes in data distributions (e.g. dynamic systems). Therefore, there is a pressing need to develop new generation deeper and robust learning models that can address these challenging learning scenarios. In this talk, I will discuss our recent work on differential graph neural networks for physics-Informed AI models via meta-learning and causal inference.
Dr. Yan Liu is a Professor in the Computer Science Department and the Director of Machine Learning Center at University of Southern California. She was a Research Staff Member at IBM Research in 2006-2010 and Chief Scientist in Didi Chuxing in 2018. She received her Ph.D. degree from Carnegie Mellon University. Her research interest is machine learning and its applications to health care, sustainability and social network analysis. She has received several awards, including ACM Distinguished Member, NSF CAREER Award, Okawa Foundation Research Award, New Voices of Academies of Science, Engineering, and Medicine, Biocom Catalyst Award Winner, and ACM Dissertation Award Honorable Mention.
Liang Zhao, firstname.lastname@example.org, Tel: (703) 993 5910
Xun Zhou, email@example.com, Tel: (319) 384-3335
Zhe Jiang, firstname.lastname@example.org, Tel: (205) 348-5243
Robert Stewart, email@example.com, Tel: (865) 574-7646