Researchers have created a new artificial intelligence (AI) tool that helps analyse how groups of people behave in cities up to 400 times faster than current available methods, according to research published in the Proceedings of the National Academy of Science journal.
The research team, led by The Alan Turing Institute’s Chief Scientist, Professor Mark Girolami, has found that the new method, which combines classical numerical models with machine learning, is faster than a previous study of the same dataset using traditional techniques.
This means that the possibility to consider large scale studies of behavioural models at city, region and indeed national scales is within our grasp.
The research reported in the paper explores examples such as how retail behaviour in London can be affected by traffic congestion policies.
The authors hope that the simplicity and speed of the method will prove a powerful tool across the quantitative sciences, such as in the social sciences, economics or computational epidemiology.
Professor Mark Girolami, The Alan Turing Institute’s Chief Scientist and paper author, said “Analysing how groups of people behave and move around in a city using observed data, such as traffic cameras, is a big challenge. But AI gives us the ability to simulate urban environments and study the behaviour and movement of people more easily.
“Our research describes one of the first ways to solve this problem using Artificial Neural Networks and we hope that this method offers a powerful tool to understand the behaviour of people cities.”
This research relates to a previous Turing research paper called Stochastic modelling of urban structure and links to the expansion of The Alan Turing Institute’s AI work.
