Presentation by Dr. Ziqi Li, University of Glasgow.  View slides from the presentation | Request to view a video recording of the presentation

Abstract

AI has demonstrated remarkable performance and has become pervasive in people's daily lives, with recent examples such as ChatGPT. AI has also achieved success in predictive tasks across a wide range of geospatial applications. However, a significant criticism of AI lies in its lack of explainability due to its inherent complexity. For geographers, explainability is often crucial as we seek to understand the underlying processes behind observed data, rather than solely focusing on prediction accuracy. Recent advancements in eXplainable AI (XAI) offer opportunities to address the explainability gap and aid in understanding complex geographical processes through the modeling of geospatial data. This talk will demostrate how XAI convergences AI with traditional spatial statistical approaches (Li, 2022) and present an empirical application of XAI in understanding travel behavior (Li, 2023). Additionally, an ongoing development of GeoShapley, a spatial XAI method, will be introduced as a work in progress. The talk will conclude by discussing the challenges associated with XAI and providing insights into future perspectives on how XAI can benefit geographical analysis.

Speaker Bio

Ziqi Li is a Lecturer/Assistant Professor of GIScience in the School of Geographical and Earth Sciences at the University of Glasgow, UK. He completed his PhD in Geography at Arizona State University in 2020. He is a Fellow of Royal Geographical Society (RGS) and serves as the Secretary of the GIScience Research Group within RGS. Ziqi has been recognized with several prestigious awards, including the Post-doctoral Enrichment Award from the Alan Turing Institute, the Nystrom Award from the American Association of Geographers (AAG), and the John Odland Award from the Spatial Analysis and Modelling Group of AAG.

Ziqi’s research focuses on the methodological development of spatially explicit and interpretable statistical and machine learning models. He is a primary developer of Multi-scale Geographically Weighted Regression (MGWR) and a core member of PySAL (Python Spatial Analysis Library). He is also broadly interested in the application of advanced spatial analysis and modeling in the fields of public health, urban analytics, travel behavior, political geography, and remote sensing and He has published 20+ articles in journals including the Annals of the American Association of Geographers, International Journal of Geographical Information Science, Computers, Environment and Urban Systems, Geographical Analysis, among others.

Lunch will be served for those in attendance. 

Register in advance to attend the meeting virtually via Zoom.
 

Dr. Li's Home Website: https://www.gla.ac.uk/schools/ges/staff/ziqili/

Dr. Li's Google Scholar: https://scholar.google.com/citations?user=g4okEdwAAAAJ

Li, Z. (2022). Extracting spatial effects from machine learning model using local interpretation method: An example of SHAP and XGBoost. Computers, Environment and Urban Systems, 96, 101845.

Li, Z. (2023). Leveraging explainable artificial intelligence and big trip data to understand factors influencing willingness to ridesharing. Travel Behaviour and Society, 31, 284-294.