Conference proceeding
Mest-GAN: Cross-City Urban Traffic Estimation with Meta Spatial-Temporal Generative Adversarial Networks
2022 IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM), pp.733-742
IEEE International Conference on Data Mining
01/01/2022
DOI: 10.1109/ICDM54844.2022.00084
Abstract
The conditional urban traffic estimation problem aims to accurately estimate the future traffic status based on the changing local travel demands, which has long been an important issue in urban planning. However, most existing methods require the target city to provide a large amount of traffic data. Once traffic estimation is performed in a "new" city where many urban services and transportation infrastructures are not built and thus no prior data is available, those works would fail due to the lack of data. In this paper, we aim to solve the conditional urban traffic estimation problem in case of data scarcity (i.e., the target city cannot provide any prior data) and tackle the main challenges including (1) knowledge learning from the source and (2) knowledge adaptation without prior traffic data. We propose a novel generative adversarial network Meta Spatial-Temporal Generative Adversarial Network (MestGAN), which can successfully estimate traffic in the target city based on local travel demands without the access to any prior traffic data. To address the first challenge, we learn the latent distribution of travel demands with the inference network, the latent distribution also indicates the diverse spatial-temporal traffic patterns. To solve the second challenge, we use the travel demand data in the target city for adaptation, where the inference network infers a latent code guiding the generator to produce accurate traffic estimations. Extensive experiments on real-world multiple-city datasets demonstrate that our Mest-GAN produces high-quality traffic estimations and outperforms state-of-the-art baseline methods.
Details
- Title: Subtitle
- Mest-GAN: Cross-City Urban Traffic Estimation with Meta Spatial-Temporal Generative Adversarial Networks
- Creators
- Yingxue Zhang - Binghamton UniversityYanhua Li - Worcester Polytechnic InstituteXun Zhou - University of IowaJun Luo - Lenovo
- Contributors
- Xingquan Zhu (Editor)Sanjay Ranka (Editor)My T Thai (Editor)Takashi Washio (Editor)Xindong Wu (Editor)
- Resource Type
- Conference proceeding
- Publication Details
- 2022 IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM), pp.733-742
- Publisher
- IEEE
- Series
- IEEE International Conference on Data Mining
- DOI
- 10.1109/ICDM54844.2022.00084
- ISSN
- 1550-4786
- eISSN
- 2374-8486
- Number of pages
- 10
- Grant note
- Safety Research using Simulation University Transportation Center (SAFER-SIM) 69A3551747131 / U.S. Department of Transportation's University Transportation Centers Program IIS-1942680; CNS-1952085; CMMI1831140; DGE-2021871 / NSF; National Science Foundation (NSF) DB210100743 / ARC Discovery Project; Australian Research Council
- Language
- English
- Date published
- 01/01/2022
- Academic Unit
- Business Analytics
- Record Identifier
- 9984419531102771
Metrics
2 Record Views