The Institute of Advanced Research in Artificial Intelligence (IARAI) is proud to announce its Traffic4cast 2020 competition. Traffic4cast is one of only a few competitions that were invited to the prestigious Neural Information Processing Systems conference (NeurIPS).
NeurIPS is the leading machine learning conference for deep learning and its competition track boasted a highly competitive field of entries.
Traffic4cast 2020 is building of its eponymous NeurIPS 2019 edition. Surprising insights from last year include:
- Our simple traffic representation as movies provided sufficient information to make state-of-the-art traffic predictions.
- Additional traffic attributes (weather, road map properties) are of value for bonus challenges and improve more complex models, albeit not markedly.
- It may be possible to separate rules underlying traffic dynamics from city-specific forecasts (transfer learning).
This year, our aim is to explore these insights further in a competition based on an order of magnitude more data, and featuring different dynamic and static attributes which we are collecting for at least 10 cities. Data describe the cities themselves, traffic, weather, and also pollution. The competition will again be made possible through IARAI’s special partnership with HERE Technologies providing these data to the scientific community. Like last year, all results are openly published in leading academic journals.