2021 Fiscal Year Annual Research Report
A Benchmark for Video-Like Urban Computing on Citywide Crowd and Traffic Prediction
| Project/Area Number |
20K19859
|
|---|
| Research Institution | The University of Tokyo |
|---|
Principal Investigator |
姜 仁河 東京大学, 情報基盤センター, 助教 (20865266)
|
|---|
| Project Period (FY) |
2020-04-01 – 2022-03-31
|
| Keywords | Spatiotemporal Data / Deep Learning / Benchmark / Traffic Prediction / Crowd Prediction / Graph Neural Networks |
|---|
| Outline of Annual Research Achievements |
By leveraging state-of-the-art deep learning technologies on spatiotemporal data, urban traffic prediction has drawn a lot of attention in AI and Intelligent Transportation System community. The problem can be uniformly modeled with a 3D tensor (T, N, C), where T denotes the total time steps, N denotes the size of the spatial domain (i.e., mesh-grids or graph-nodes), and C denotes the channels of information.
In this year, we first synthetically review the deep traffic models as well as the widely used datasets, then build a standard benchmark to comprehensively evaluate their performances with the same settings and metrics. Our study named DL-Traff is implemented with two most popular deep learning frameworks, i.e., TensorFlow and PyTorch, which is already publicly available as two GitHub repositories, https://github.com/deepkashiwa20/DL-Traff-Grid and https://github.com/deepkashiwa20/DL-Traff-Graph.
Furthermore, based on this benchmark, we have also developed a series of deep learning models for some spatiotemporal forecasting tasks (e.g, ambulance demand, city incident occurance) by utilizing the state-of-the-art Graph Neural Networks (GNNs). The relevant research achievements have been published at top international AI or Data Science conferences including ICDE 2021 and CIKM 2021. In particular, our benchmark DL-Traff was awarded CIKM 2021 Best Resource Paper Runner-Up.
|
Research Products
(9 results)
