PredPol Brings Artificial Neural Networks To Law Enforcement?
February 18, 2013 in Technology
The futuristic world of “Minority Report” has become a fantasy for many technology companies. One of the more controversial technologies shown in the movie is the ability to see the future by holding psychic beings captive and harnessing their abilities, a practice that Tom Cruise (the character’s name isn’t important) endeavors to end.
PredPol, a cloud-based software company, has raised $1.3 million from Plantronics’ Ken Kannappan and Barbara Scherer and a handful of other investors in a friends and family seed round to help make the Minority Report future a reality – with one key difference. “We don’t have anyone laying in baths predicting the future,” as PredPol’s director of government relations and strategy and former mayor of Santa Cruz Ryan Coonerty puts it.
That’s a relief.
PredPol began as an academic study by UCLA professors Jeff Brantingham and George Mohler. (Mohler has since moved to Santa Clara University.) Originally used to predict earthquakes, Brantingham and Mohler tweaked the algorithms they had developed to predict crime. Law enforcement agencies that license PredPol’s software via a SaaS model are able to collect data about criminal acts and where they occur; the software then uses that data and predicts where and when future crimes are most likely to occur, down to 500 square foot blocks.
Developed via a research partnership with the Los Angeles Police Department, the professors’ algorithm has been used on a trial basis since 2011. CEO Caleb Baskin says that PredPol’s software became twice as effective during that trial period, and the LAPD saw a 36 percent drop in crime rates in the participating district.
A for-profit company with an exclusive license on the professors’ work from UCLA and Santa Clara University, PredPol’s cloud-based SaaS offering is already used by six law enforcement agencies in the US. The company is currently in talks with six more – one of which is overseas – and that number can be expected to grow; some 225 agencies have approached PredPol seeking to utilize the company’s technology in some form.
Of course, PredPol is continually working on its core product to improve accuracy and effectiveness. I asked Coonerty if the software might someday become sophisticated enough to predict crime on an individual level, and he was quick to set me straight. “We have no plans to move anywhere close to the individual. You’ll – rightly – run into a variety of Constitutional problems almost immediately.” If anything, he says, the company is helping remove bias against individuals from the equation.
“If you send police officers out with a mandate to cut crime by 10 percent by the end of the quarter, they’re going to use their gut instincts to do it. [PredPol] gives them more knowledge and allows them to focus on problem places at the right time instead of focusing on individuals,” he says.
PredPol is helping law enforcement agencies in two ways, then. The first is by telling them how to better use their limited resources through big data and modern technologies. The second is preventing crime, which helps the community (in obvious ways) but also saves law enforcement agencies money.
“[PredPol is a] new but well-backed company that is bringing private sector analytics to the public sector, [specifically] law enforcement agencies,” Baskin says. “They need to find ways to do more with less. A tool that tells them how to allocate resources to fight crime in places before they occur fits the bill.”
Neural Network Design in Cloud Computing (SAAS) Software As A Service Model
Brain waves encode rules for behavior.