In Transport and Telecommunication Institute continues implementation of the project 220.127.116.11/VIAA/1/16/112 «Spatiotemporal urban traffic modelling using big data» (postdoc: Dr.sc.ing. Dmitry Pavlyuk). The project is supported by the European Regional Development Fund (ERDF), within the 18.104.22.168 Post-doctoral Research Aid activity.
During the During the first quarter of 2019, we executed several interesting activities:
- Developed software for automated collection and preprocessing of traffic-related data. Source codes of developed functions are available in public repository https://github.com/DmitryPavlyuk/postdoc
- Executed a research on application of video prediction algorithms and models for spatiotemporal urban traffic forecasting. Video stream forecasting is an emerging area of modern AI models’ applications, which widely used in self-driving vehicles. We suggested a direct application of existing video prediction models for spatiotemporal urban traffic forecasting and supported this suggestion by a real-world case -study. Research results were described in the paper “Spatiotemporal traffic forecasting as a video inpainting problem”, which is accepted for publication in Xplore database and will be presented at the 6th International Conference on Models and Technologies for Intelligent Transportation Systems (MT-ITS), 5-7 June 2019, Cracow, Poland.
- Started research of ensemble learning for spatiotemporal feature selection in traffic flows. Ensemble learning is a scientific representation of the proverb “two heads are better than one”; in our case it simultaneously applies several feature selection methods (and a combination procedure) for more efficient, robust and parsimonious models. At this moment, the resulting paper «Towards ensemble learning of traffic flows’ spatiotemporal structure» is under review for tan important transport conference – the 22nd Euro Working Group on Transportation Meeting (EGWT), 18-20 September 2019, Barcelona, Spain
Current information on the progress and results of the project is available here.