Science of Machine Learning

Where are hotspots of "2020" traffic ?

- Predict the traffic of future events based on the individual behavior -

Abstract

Because an increase of “future events” that are events not organized in the past like the 2020 Olympic and Paralympic is possible, communicating a large amount of “future events” traffic efficiently with appropriate quality is a challenge. A conventional method of predicting a event traffic is for regularly-scheduled events. Thus, applying this method to future events is difficult. We have studied a spatio-temporal traffic prediction based on models of event-specific communication behavior combined with human mobility. Furthermore, focusing on purposes of human behavior are aggregated by event types, we have developed an efficient traffic prediction technology by using multi-agent simulation.

Photos

Poster


Please click the thumbnail image to open the full-size PDF file.

Presenters

 Hiroyuki Toda
Akihiro Shiodu
Network Technology Laboratories
 Hiroyuki Toda
Yasuhiro Ikeda
Network Technology Laboratories
 Hiroyuki Toda
Shouji Kouno
Network Technology Laboratories
Kouhei Shiomoto
Kouhei Shiomoto
Machine Learning and Data Science Center