Trending Topics
Sponsored Content

Searching for fugitives: How machine learning can help you find them

Digital investigation technology can help turn up leads when traditional investigations stall

Sponsored by
CCTV Cameras and Big Ben London Landmark

Cellebrite Pathfinder and the city’s network of CCTV cameras have helped solve a murder in London, among other difficult cases.

JohnDWilliams/Getty Images/iStockphoto

When a police officer pulls over a speeding driver, they generally have all the evidence they require for a citation and, if it’s contested, a conviction. Their radar readings, any videos and their eyewitness testimony are sufficient in most cases to establish the crime was committed and the vehicle operator was at fault.

With each stop, however – even given a willingly presented driver’s license, registration and proof of insurance – officers will also run the driver’s name and tag number through their databases of the wanted. Obviously, this is to ensure the driver isn’t being sought for, and the car hasn’t been associated with, something worse than speeding. And this is entirely legal and broadly accepted in the U.S.: When someone’s unlawful activity comes to the attention of law enforcement, it isn’t just “fishing” to check into who they are – and aren’t.

These days, a lot more than names and warrants can be turned up on criminal suspects by investigators using modern tools. And while it might not be feasible to employ such technologies at every roadside traffic stop, they can be useful when bigger and more serious investigations bog down, helping locate individuals, leads and evidence that are harder to unearth with traditional methods.

Accessing the deeply buried layers of digital evidence hidden within suspects’ smartphones and mobile devices and across our increasingly interconnected society isn’t as hard as you might imagine – and can yield game-changing results.

The Pathfinder platform from Cellebrite uses machine learning to help unearth, organize, categorize, map and streamline such complex data across a broad range of sources – including apps, browsers, cell towers and Wi-Fi hotspots – to illuminate subjects, connections and activities that can remain elusive without such a sophisticated technological assist.

A pair of company experts discussed the various ways “casting a net” with an asset like Pathfinder can help turn up new information and reinvigorate stalled investigations in a recent webinar. When all else fails, they emphasized, you can leverage technology to pursue leads for you.

“Sometimes you might catch a shoe,” said Cellebrite’s Sigal Caspi, who served 25 years with the Israeli police, “but sometimes you can also get the dollars.”


An easy place to start is with the backlogs of unclosed cases that plague many departments. These may loiter unresolved not because they’re especially difficult or sophisticated, but simply for reasons of investigative manpower. Fraud cases, as one example, may be prevalent but a lower priority, pushed to the back burner by killings, assaults and other high-priority cases.

Meanwhile, departments are constantly receiving red notices from Interpol and bulletins from other partners about important fugitives. Perhaps their wanted fraudster is your wanted fraudster: With a simple click and drag, the Pathfinder platform lets investigators integrate information from those red notices into their open fraud case, where it can search for similarities and links.

One example involved a Virginia gang shooting that mistakenly killed a 1-year-old child. Investigators worked the case furiously for more than a year, ultimately amassing more than 50,000 images. But that was too many to review manually and therefore didn’t produce much of value, though they did elicit the nickname of a potential person of interest.

Then detectives did a search on that nickname using Pathfinder’s optical character recognition (OCR) abilities to rapidly scan and convert text from images or documents into digital formats. From that mountain of images, the OCR search excavated a hit, finding the nickname in connection with a damning photo of the suspects casing the apartment where the shooting was planned. Additional images showed them climbing the residence’s stairs and turning toward the door.

The discovery was enough to restart the case.

“Nobody is going to sit down and scroll through 50,000 images and read texts line by line to determine whether they’re important,” said Jennifer Gudaitis, a solutions engineer with Cellebrite who spent 14 years in law enforcement, including working as a homicide sergeant. “In this case, [OCR] did that work for us. It went through and read all the words in that dataset and found that nickname for me. It was months and months and months of this investigation. And with the keyword search, it came back in less than 30 seconds.”

“It may seem odd to preserve evidence of your crime on your phone, but people tend to save what’s important and ‘close to their heart’,” Caspi added. A phone’s gallery is a good place for any investigator to begin.


Caspi provided another example of the OCR function’s value. In this case a man was found dead in a London hotel room, a phone tossed under the bed. Investigators had little to go on, but they used Pathfinder to search the phone for the word “camera.” That unearthed a tense text exchange about a closed-circuit TV camera one party had apparently missed in performing the crime. An OCR search also found an incriminating exchange about that error.

As most people know, London is one of the most surveilled cities in the world, with more than 940,000 CCTV cameras citywide. Police checked cameras around the crime scene and discovered an image of the victim with another man. A quick click on “adjacent events,” where Pathfinder organized everything else it turned up around the time and place recorded in the image’s metadata, revealed a fuller exchange where the assassins realized their critical mistake.

“In the beginning they were very happy the job was done,” recalled Caspi. “But as we go farther along, they start to understand they have a problem. There was a CCTV camera that was overlooked … And actually, you get the narrative of the crime in a nutshell.”


In many ways, the move to personal electronic devices and communications throughout society has been a benefit to law enforcement. The ubiquity of our phones in our lives – not to mention the prevalence of cameras, hotspots and other information-rich sources around us – helps create reconstructable tales of our activities and associations. Where police may be limited is in the time and expertise needed to dig out the information of value from the range of sources that contain it, then interpret and connect it all for successful prosecution.

With Pathfinder, Cellebrite offers an organizational solution that works as a partner for law enforcement, reducing both the time and technical know-how required to extract data important to investigations and winnow it down to the pieces that make a difference. This can help provide fast and definitive resolutions to some of today’s most difficult cases.

As Gudaitis said, “You can’t escape technology.”

For more information, visit Cellebrite Pathfinder.

John Erich is a Branded Content Project Lead for Lexipol. He is a career writer and editor with more than two decades of experience covering public safety and emergency response.