A method to evaluate task-specific importance of spatio-temporal units based on explainable artificial intelligence

ABSTRACT

Big geo-data are often aggregated according to spatio-temporal units for analyzing human activities and urban environments. Many applications categorize such data into groups and compare the characteristics across groups. The intergroup differences vary with spatio-temporal units, and the essential is to identify the spatio-temporal units with apparently different data characteristics. However, spatio-temporal dependence, data variety, and the complexity of tasks impede an effective unit assessment. Inspired by the applications to extract critical image components based on explainable artificial intelligence (XAI), we propose a spatio-temporal layer-wise relevance propagation method to assess spatio-temporal units as a general solution. The method organizes input data into an extensible three-dimensional tensor form. We provide two means of labeling the spatio-temporal tensor data for typical geographical applications, using temporally or spatially relevant information. Neural network training proceeds to extract the global and local characteristics of data for corresponding analytical tasks. Then the method propagates classification results backward into units as obtained task-specific importance. A case study with taxi trajectory data in Beijing validates the method. The results prove that the proposed method can evaluate the task-specific importance of spatio-temporal units with dependence. This study also attempts to discover task-related knowledge using XAI.

KEYWORDS:

• Deep learning
• tensor
• spatio-temporal dependence
• task-specific
• classification

Acknowledgments

The authors would like to thank Prof. David O’Sullivan, Prof. May Yuan, and the anonymous reviewers for their comments, and Prof. Shaowen Wang, Dr. Fan Zhang, Dr. Xuexi Yang, Dr. Di Zhu, Miss. Xiaoyue Xing and the members of CyberGIS Center for their valuable advice.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. Government notice: http://www.gov.cn/zhengce/content/2015-12/10/content_10394.htm

2. TensorFlow LRP Wrapper: https://github.com/VigneshSrinivasan10/interprettensor

Additional information

Funding

This work is supported by the National Natural Science Foundation of China [grant numbers 41625003, 41830645, 41771425, and 41571397];

Notes on contributors

Ximeng Cheng

Ximeng Cheng received B.S. and M.S. degrees from China University of Geosciences, Beijing, China, in 2013 and 2016, respectively. He received a Ph.D. degree in cartography and GIS from Peking University, Beijing, China, in 2020. His research interests include GIScience, spatio-temporal data mining, GeoAI, and urban studies.

Jianying Wang

Jianying Wang received a B.S. degree from Surveying and Geo-informatics College, Tongji University in 2016. He is currently pursuing a Ph.D. degree in GIScience with the Institute of Remote Sensing and Geographical Information Systems, Peking University. His primary research interest lies in spatial data mining and transport geography.

Haifeng Li

Haifeng Li is a Professor with the School of Geosciences and Info-Physics, Central South University, Changsha, China. He received a master degree from the South China University of Technology, and a Ph.D. degree in photogrammetry and remote sensing from Wuhan University. His current research interests include geo/remote sensing big data, machine/deep learning, and artificial/brain-inspired intelligence.

Yi Zhang

Yi Zhang is an Associate Professor in the Institute of Remote Sensing and Geographical Information Systems, Peking University, with a Ph.D. degree in cartography and geography information system. He has approximately seventeen years of professional and academic experience in the field of geo-information, spatial analysis, big geo-data and social sensing.

Lun Wu

Lun Wu is Professor of GIScience in the Institute of Remote Sensing and Geographical Information Systems, Peking University. He received his B.S., M.S., and Ph.D. degrees in geography from Peking University, China, in 1984, 1987 and 1990, respectively. His research interests cover several theoretical and technical aspects in big geo-data science and smart city applications.

Yu Liu

Yu Liu is currently the Boya Professor of GIScience at the Institute of Remote Sensing and Geographical Information Systems, Peking University. He received his B.S., M.S., and Ph.D. degrees from Peking University in 1994, 1997, and 2003, respectively. His research interest mainly concentrates in humanities and social science based on big geo-data.