Fujitsu’s AI Tech Will Help Catch Escaping Criminals

Japanese tech-giant Fujitsu, along with the University of Electro-Communications, has figured out a way to catch criminals on the run using artificial intelligence technology.

The teams developed a high-speed algorithm that uses mathematical game theory as an artificial intelligence technology to assist with security planning. The technology will help solve city-scale road network security problems, such as where best to position checkpoints when trying to catch a criminal.

When people gather in large places, like city streets or airports, it would be ideal if all paths could be sealed off in the event of emergencies or as a precaution. However this is a difficult task to achieve due to limited security resources.

In recent years, there has been more focus on the use of AI to formulate security. For example, technology using game theory, which treats both the criminal's side and the security side as opposing decision makers, called security games, is beginning to come into practical use as a tool to help experts make decisions.

However it has been difficult to apply game theory to a city-scale security problem of catching criminals at checkpoints in real-world cities because the processing volume expands exponentially with the scale of the road network.

But now, using Fujitsu Laboratories' network contraction technology, the team created an algorithm to quickly solve city-scale road network security problems. Compared with previous technology, this makes it possible to find the hypothetically optimal security plan 20 times faster, on average, for a 100-node problem, and 500 times faster, on average, for a 200-node problem.

In a place like Tokyo, which could see a 200,000-node problem, a plan could take days to come up with, but the new technology could generate a security plan in about five minutes.

Fujitsu Laboratories hopes to commercialize this technology next year, while the University of Electro-Communications expands on it beyond city-scale road networks.