UCL News


Real-life "Minority Reports" to help police reduce crime

26 March 2003

Scientists will today launch a new computer technique able to predict where and when crime might take place, enabling the police to locate crime before it happens.

Jonathan Corcoran and Professor Andrew Ware (School of Computing, University of Glamorgan) will present a new crime mapping technique modelled on the human brain which has the ability to make accurate and reliable predictions about where and when crimes will occur for the first time.

Their prediction is made using a relatively new branch of computing, a form of artificial intelligence called neural networks, which replicate the way in which the brain learns from experience. Neural networks can analyse the relationships within sets of data and 'learn' what the future might hold from past events.

The neural networks have been 'taught' using information on past crimes and, once trained, have the ability to make predictions about criminal activities within a town. The neural networks concentrate on crime "hot-spots" and are able to provide the police with a profile that predicts what type of criminal activity is likely to occur and at what time.

Researcher Jonathan Corcoran said: "This new technique will allow the police to make the best of limited resources, for example deploying police to high-risk areas at just the right time to help in the prevention of likely criminal activity."

UCL's Jill Dando Professor of Crime Science, Gloria Laycock, said: "Crime mapping techniques like this are powerful new tools in the fight against crime. This technique is just one of the many ways in which crime science is helping to reduce crime across the UK."