Traffic4cast at NeurIPS 2020 - yet more on the unreasonable effectiveness of gridded geo-spatial processes

Michael Kopp, David Kreil, Moritz Neun, David Jonietz, Henry Martin, Pedro Herruzo, Aleksandra Gruca, Ali Soleymani, Fanyou Wu, Yang Liu, Jingwei Xu, Jianjin Zhang, Jay Santokhi, Alabi Bojesomo, Hasan Al Marzouqi, Panos Liatsis, Pak Hay Kwok, Qi Qi, Sepp Hochreiter
Proceedings of the NeurIPS 2020 Competition and Demonstration Track, PMLR 133:325-343, 2021.

Abstract

The IARAI Traffic4cast competition at NeurIPS 2019 showed that neural networks can successfully predict future traffic conditions 15 minutes into the future on simply aggregated GPS probe data in time and space bins, thus interpreting the challenge of forecasting traffic conditions as a movie completion task. U-nets proved to be the winning architecture then, demonstrating an ability to extract relevant features in the complex, real-world, geo-spatial process that is traffic derived from a large data set. The IARAI Traffic4cast challenge at NeurIPS 2020 build on the insights of the previous year and sought to both challenge some assumptions inherent in our 2019 competition design and explore how far this neural network technique can be pushed. We found that the prediction horizon can be extended successfully to 60 minutes into the future, that there is further evidence that traffic depends more on recent dynamics than on the additional static or dynamic location specific data provided and that a reasonable starting point when exploring a general aggregated geo-spatial process in time and space is a U-net architecture.

Cite this Paper

BibTeX
@InProceedings{pmlr-v133-kopp21a, title = {Traffic4cast at NeurIPS 2020 - yet more on the unreasonable effectiveness of gridded geo-spatial processes}, author = {Kopp, Michael and Kreil, David and Neun, Moritz and Jonietz, David and Martin, Henry and Herruzo, Pedro and Gruca, Aleksandra and Soleymani, Ali and Wu, Fanyou and Liu, Yang and Xu, Jingwei and Zhang, Jianjin and Santokhi, Jay and Bojesomo, Alabi and Marzouqi, Hasan Al and Liatsis, Panos and Kwok, Pak Hay and Qi, Qi and Hochreiter, Sepp}, booktitle = {Proceedings of the NeurIPS 2020 Competition and Demonstration Track}, pages = {325--343}, year = {2021}, editor = {Escalante, Hugo Jair and Hofmann, Katja}, volume = {133}, series = {Proceedings of Machine Learning Research}, month = {06--12 Dec}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v133/kopp21a/kopp21a.pdf}, url = {https://proceedings.mlr.press/v133/kopp21a.html}, abstract = {The IARAI Traffic4cast competition at NeurIPS 2019 showed that neural networks can successfully predict future traffic conditions 15 minutes into the future on simply aggregated GPS probe data in time and space bins, thus interpreting the challenge of forecasting traffic conditions as a movie completion task. U-nets proved to be the winning architecture then, demonstrating an ability to extract relevant features in the complex, real-world, geo-spatial process that is traffic derived from a large data set. The IARAI Traffic4cast challenge at NeurIPS 2020 build on the insights of the previous year and sought to both challenge some assumptions inherent in our 2019 competition design and explore how far this neural network technique can be pushed. We found that the prediction horizon can be extended successfully to 60 minutes into the future, that there is further evidence that traffic depends more on recent dynamics than on the additional static or dynamic location specific data provided and that a reasonable starting point when exploring a general aggregated geo-spatial process in time and space is a U-net architecture.} }
Endnote
%0 Conference Paper %T Traffic4cast at NeurIPS 2020 - yet more on the unreasonable effectiveness of gridded geo-spatial processes %A Michael Kopp %A David Kreil %A Moritz Neun %A David Jonietz %A Henry Martin %A Pedro Herruzo %A Aleksandra Gruca %A Ali Soleymani %A Fanyou Wu %A Yang Liu %A Jingwei Xu %A Jianjin Zhang %A Jay Santokhi %A Alabi Bojesomo %A Hasan Al Marzouqi %A Panos Liatsis %A Pak Hay Kwok %A Qi Qi %A Sepp Hochreiter %B Proceedings of the NeurIPS 2020 Competition and Demonstration Track %C Proceedings of Machine Learning Research %D 2021 %E Hugo Jair Escalante %E Katja Hofmann %F pmlr-v133-kopp21a %I PMLR %P 325--343 %U https://proceedings.mlr.press/v133/kopp21a.html %V 133 %X The IARAI Traffic4cast competition at NeurIPS 2019 showed that neural networks can successfully predict future traffic conditions 15 minutes into the future on simply aggregated GPS probe data in time and space bins, thus interpreting the challenge of forecasting traffic conditions as a movie completion task. U-nets proved to be the winning architecture then, demonstrating an ability to extract relevant features in the complex, real-world, geo-spatial process that is traffic derived from a large data set. The IARAI Traffic4cast challenge at NeurIPS 2020 build on the insights of the previous year and sought to both challenge some assumptions inherent in our 2019 competition design and explore how far this neural network technique can be pushed. We found that the prediction horizon can be extended successfully to 60 minutes into the future, that there is further evidence that traffic depends more on recent dynamics than on the additional static or dynamic location specific data provided and that a reasonable starting point when exploring a general aggregated geo-spatial process in time and space is a U-net architecture.
APA
Kopp, M., Kreil, D., Neun, M., Jonietz, D., Martin, H., Herruzo, P., Gruca, A., Soleymani, A., Wu, F., Liu, Y., Xu, J., Zhang, J., Santokhi, J., Bojesomo, A., Marzouqi, H.A., Liatsis, P., Kwok, P.H., Qi, Q. & Hochreiter, S.. (2021). Traffic4cast at NeurIPS 2020 - yet more on the unreasonable effectiveness of gridded geo-spatial processes. Proceedings of the NeurIPS 2020 Competition and Demonstration Track, in Proceedings of Machine Learning Research 133:325-343 Available from https://proceedings.mlr.press/v133/kopp21a.html.

Related Material