Cyber detective links up crimes

December 02, 2004

MANY more crimes might be solved if detectives were able to compare the records for cases with all the files on past crimes. Now an artificial intelligence system has been designed to do precisely that. Working 24 hours a day, seven days a week, it could look for telltale similarities in crime records and alert detectives when it finds them. Developed by computer scientists Tom Muscarello and Kamal Dahbur at DePaul University in Chicago, the system uses pattern-recognition software to link related crimes that may have taken place in widely separated areas whose police forces may rarely be in close contact. Called the Classification System for Serial Criminal Patterns (CSSCP), the system sifts through all the case records available to it, assigning numerical values to different aspects of each crime, such as the kind of offence, the perpetrator's sex, height and age, and the type of weapon or getaway vehicle used. From these figures it builds a crime description profile.

A neural network program then uses this to seek out crimes with similar profiles. The neural network the DePaul team uses, called a Kohonen network, is particularly good at finding patterns in a set of input data without any human intervention, Muscarello says. Some neural networks require an operator to "train" them to find patterns in data sets - but this requires foreknowledge of the pattern. If it finds a possible link between two crimes, CSSCP compares when and where they took place to find out whether the same criminals would have had enough time to travel from one crime scene to the other. In a ring-fenced lab trial of the system using three years' worth of data on armed robbery- it wasn't operating on a live police network- Muscarello claims the system was able to spot 10 times as many patterns as a team of detectives with access to the same data. Muscarello stresses that CSSCP will not replace human detectives. It simply provides a starting point for detectives by flagging potentially related crimes. While detectives could probably identify the same patterns, the sheer mass of available data usually makes this too time-consuming. In the UK an online version of a manually searchable crime database called Crimelink was launched this week. But while Crimelink can be used to search for specified patterns of criminal activity- querying if, say, a certain type of car has been used in armed robberies in Cambridge- it will not search for similarities that have not been suggested by a human operator.

Muscarello is now hoping to convince the Chicago police department to run trials of the system. Feedback from six leading detectives in the Chicago area have been positive, and he says one reason for this is that the researchers consulted detectives on how the system should work. "To a certain extent we based it on the way detectives solve these crimes," Muscarello says. In conventional investigations, the work of finding a pattern may be split between several detectives, with each one handling a different aspect of the case. For example, one might focus on the witnesses or victims, while another might try to trace a getaway vehicle. CSSCP can do the same by applying different neural networks to find patterns in different facets of the investigation. The amount of information being filed for each crime is constantly increasing, says John Kingston, director of the Joseph Bell Centre for Forensic Statistics and Legal Reasoning at the University of Edinburgh, UK. This means that "data mining" is likely to become increasingly necessary in crimefighting, he predicts.

This article appears in New Scientist issue: 4 December 2004

PLEASE MENTION NEW SCIENTIST AS THE SOURCE OF THIS STORY AND, IF PUBLISHING ONLINE, PLEASE CARRY A HYPERLINK TO: "These articles are posted on this site to give advance access to other authorised media who may wish to quote extracts as part of fair dealing with this copyrighted material. Full attribution is required, and if publishing online a link to is also required. Advance permission is required before any and every reproduction of each article in full - please contact Please note that all material is copyright of Reed Business Information Limited and we reserve the right to take such action as we consider appropriate to protect such copyright."

New Scientist

Related Neural Networks Articles from Brightsurf:

Deep neural networks show promise for predicting future self-harm based on clinical notes
Medical University of South Carolina researchers report in JMIR Medical Informatics that they have developed deep learning models to predict intentional self-harm based on information in clinical notes.

Researchers develop new model of the brain's real-life neural networks
Researchers at the Cyber-Physical Systems Group at the USC Viterbi School of Engineering, in conjunction with the University of Illinois at Urbana-Champaign, have developed a new model of how information deep in the brain could flow from one network to another and how these neuronal network clusters self-optimize over time.

The brain's memory abilities inspire AI experts in making neural networks less 'forgetful'
Artificial intelligence (AI) experts at the University of Massachusetts Amherst and the Baylor College of Medicine report that they have successfully addressed what they call a ''major, long-standing obstacle to increasing AI capabilities'' by drawing inspiration from a human brain memory mechanism known as ''replay.''

New data processing module makes deep neural networks smarter
Artificial intelligence researchers have improved the performance of deep neural networks by combining feature normalization and feature attention modules into a single module that they call attentive normalization.

Neural cartography
A new x-ray microscopy technique could help accelerate efforts to map neural circuits and ultimately the brain itself.

Researchers study why neural networks are efficient in their predictions
A study has tested the predictions of a neural network to check whether they coincide with actual results.

Optimizing neural networks on a brain-inspired computer
Neural networks in both biological settings and artificial intelligence distribute computation across their neurons to solve complex tasks.

Teaching physics to neural networks removes 'chaos blindness'
Teaching physics to neural networks enables those networks to better adapt to chaos within their environment.

A clique away from more efficient networks
An old branch of mathematics finds a fertile new field of application.

Unravelling complex brain networks with automated 3D neural mapping
KAIST researchers developed a new algorithm for brain imaging data analysis that enables the precise and quantitative mapping of complex neural circuits onto a standardized 3D reference atlas.

Read More: Neural Networks News and Neural Networks Current Events is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to