- Date : Wednesday, August 5, 2026
- Time : 3:00PM – 4:00PM (KST, Seoul Time)
- Venue : Zoom (Online Research Seminar)
- Speaker : Prof. Shenhao Wang, University of Florida
Artificial Intelligence for Travel Demand Modeling
Abstract
Travel demand modeling has long relied on discrete choice models (DCMs) grounded in random utility theory, but the rise of artificial intelligence now offers powerful new tools for the field. This talk presents a research program that synergizes deep learning with classical choice modeling across three studies. The first benchmarks over 7,000 experiments across more than 100 machine learning and discrete choice models, showing that although AI models often outperform traditional DCMs, predictive performance is influenced more by data characteristics than model categories. The second introduces deep hybrid models that combine survey data with satellite imagery, improving prediction while revealing interpretable spatial patterns and enabling counterfactual urban imagery generation. The third develops graph neural network-based choice models for residential location choice, capturing spatial correlations among alternatives and outperforming classical logit and feedforward models. Together, these studies illustrate how AI can enrich classical travel demand modeling, pointing toward next-generation models that are more predictive, interpretable, and multimodal across urban contexts.
About the Speaker
Dr. Shenhao Wang is an assistant professor and director of the Urban AI Laboratory at the University of Florida. His research develops artificial intelligence foundations for urban and transportation systems along three themes: individual travel behavior, complex mobility systems, and generative planning practices. His work has been funded by the Department of Energy, the Singapore-MIT Alliance for Research and Technology, the National Science Foundation, and industry partners. Dr. Wang completed his interdisciplinary Ph.D. in Computer and Urban Science at the Massachusetts Institute of Technology in 2020, where he also earned Master of Science in Transportation and Master of City Planning degrees.
How to Join
This is an open online seminar hosted by KAIST IMPACTs. All are welcome to join over Zoom.
Zoom Meeting ID : 950 3606 8654
Passcode : 792397
For further information, please contact Dr. Fan Wu (fanwu@kaist.ac.kr).
