Exhibition Program

Science of Machine Learning

03

Recover urban people flow from population data

- People flow estimation from spatiotemporal population data -

Abstract

Real-time spatiotemporal population data is attracting a great deal of attention for understanding crowd movements in cities. The data is the aggregation of personal location information and consists of just areas and the number of people in each area at certain time instants. Accordingly, it does not explicitly represent crowd movement. We propose a probabilistic collective graphical models that can estimate crowd movement from spatiotemporal population data. There are two technical challenges: (i) poor estimation accuracy as the traditional approach means the model would have too many degrees of freedom, (ii) excessive computation cost. Our key idea is to model the transition probability between areas by using three factors: departure probability of areas, gathering score of areas, and geographical distance between areas. These advances enable us to reduce the degrees of freedom of the model appropriately and derive an efficient estimation algorithm.

References

  • [1] Y. Akagi, T. Nishimura, T. Kurashima, H. Toda, “A fast and accurate method for estimating people flow from spatiotemporal population data,” in Proc. the 27th International Joint Conference on Artificial Intelligence and the 23rd European Conference on Artificial Intelligence (IJCAI-ECAI-18), pp. 3293-3300, 2018.

Poster

Photos

Contact

Yasunori Akagi, Proactive Navigation Project, NTT Service Evolution Laboratories
Email: