Theft motivates crime victim into becoming crime fighter

While on a visit to Bologna, Italy in 2014, Mami Kajita fell victim to a pickpocket.

"It was around 1 p.m. on a Sunday and I realized my wallet had been stolen," she relates. "Reporting it to the police, I was told, 'Around that time of day most local people are at home, which means there are more tourists about.'"

Kajita's loss proved transformative and she set out to harness her technological knowhow to fight crime.

Fast forward three years to 2017, when Kajita, then a visiting researcher at the University of Tokyo, established a corporation with the unusual name of Singular Perturbations Inc. We know, it's a bit of a mouthful, so like Weekly Playboy (Sept 9), we'll refer to it herein as SP for short.

SP's main service is a crime forecasting system named "Crime NABI." (Nabi is taken from the English navigation.) Last year, SP established an office in Brazil's largest city of Sao Paulo and is currently tied up with 11 government organizations in Central and Latin America.

"Thefts of copper electrical cables have been a frequently occurring problem in Brazil," explains Kajita, whose title in the organization is CEO. "This poses a major challenge in maintaining infrastructure. After our system underwent two months of tests in cooperation with the police in Belo Oriente, a municipality in southeast Brazil, authorities confirmed that incidences of cable thefts were down by 69%. So the local government decided on full-scale adoption of the Crime NABI system."

From childhood Kajita liked studying the natural sciences, and felt that natural phenomena could be understood through mathematics. After obtaining her doctorate from Todai in statistical mechanics, Kajita set up Singular Perturbations, taken from the term for type of research method first developed in the United States in the 1940s.

Her Crime NABI system harnesses two types of algorithms to analyze a variety of data, including previous crimes, population statistics, land usage, types of buildings and thoroughfares, satellite images, weather and other factors, to forecast crimes.

"The first forecasting involves chronological data," Kajita explains. "When criminals are successful, they tend to repeat their crimes. The pattern of past crimes is analyzed. The other algorithm involves spatial data. The pattern of crimes is analyzed according to such factors as population density and geographic data.

"Since this generates a huge volume of data, the key issue is how to deal with the calculation time and higher cost that computer processing would entail. We also developed proprietary algorithms for data compression, which has achieved the desired high speed while realizing low costs.

"Even in situations where data is insufficient, or even for areas where crime is practically nonexistent, we can provide highly accurate forecasts by using a mechanism called transference, a type of machine learning," Kajita said.

"In societies where firearms are widespread, police lives are at risk merely by going out on patrol, as in high-crime areas many police killed in the line of duty," Kajita tells the magazine. "So it will be important to set up patrol routes in real time and suggest locations for installing surveillance cameras.

In addition to purposes aside from maintenance of public order, SP has been harnessed in Brazil to track thefts of petroleum products, another common crime.

"Using Crime NABI, safer transport routes with lower incidences of theft can be selected, which also enables reduction in insurance rates.

"This is why many of our clients are in the insurance, security and travel business sectors," she said.

Fortunately or unfortunately, incidence of crimes in Japan is too low to justify adoption of the Crime NABI system here. But Kajita is thinking to put her knowhow to use to create an application for fellow Japanese who travel abroad.

"Being victimized by a pickpocket eventually led to my coming up with the idea for this business," she pointed out. "I can envisage a product where users could ask, 'Will it be safe for me to walk around such-and-such a place at this particular time of day?'

"If such an app helps even one person avoid an unhappy experience, that would make me happy," she said.

© Japan Today