Security Systems Next Leap Behavioral Analytics

(Newswire.net — September 15, 2014) Huntingdon Valley, PA — Ray Kurzweil is described by Bill Gates as one of the brightest people on the planet. Presently Kurzweil is director of engineering at Google, and he holds basic patents in computer speech recognition, text-to-speech synthesis and optical character recognition. Forbes Magazine described him as “the ultimate thinking machine.” But it is his futurist prediction of an approaching time when computers will be as intelligent as we are, which he calls “the singularity” that causes the most attention and controversy. He is a believer in “strong A.I.” (artificial intelligence) meaning that he predicts that computers will be self-conscious, just as we are, and will then become a trillion times as conscious as we are. His book “The Singularity is Near” presents his theories in fascinating detail.

 

Here is how this relates to security. The singularity just got nearer. Video surveillance has advanced from simple cameras to artificial intelligence.

 

 

  

 

The mind of a thinking machine

 

 

The Pace of Evolution of Artificial Intelligence in Security is Stunning

 

Initially video surveillance camera systems were installed with the idea that a security guard would watch a bank of monitors on a wall. He or she was supposed to stare at the thirty-two or forty-eight camera monitors all day long. That proved to be an impractical idea. The next advance was motion-detecting cameras. They would only activate when there was motion. Indoors, after hours, that is practical. But outdoors things constantly move like trees blowing in the wind, dogs, birds, clouds, headlights. So that caused hundreds or thousands of false alerts per camera per day. Then a major advance was made utilizing artificial intelligence. Now the system could recognize the human form distinct from say a big, dog or a deer or anything else.

 

Rules could be set and the system had the intelligence to know when they were broken. For example no one is allowed in a certain area during certain times, or a vehicle is allowed to drive in one direction on a driveway but not the other. It’s not okay for this particular object to be moved but that one is okay. This is called “rule based video analytics.” So the computer artificial intelligence watched all the cameras all the time, without “blinking an eye” or getting tired or falling asleep. It could see and think like a human guard, only much better in some ways. For example in rain, glare, fog, at a great distance it could distinguish an intruder where a human guard would never be able to spot them. The alert was then sent to a security person or monitoring station and they could immediately talk down over public address speakers to the intruder in real time.

 

That was the present state of the technology and was used by Homeland Security and major airports, ports, and so forth and went into use for ordinary businesses, schools, hospitals, car dealerships, chemical plants, warehouses and even high end residential estates. 

 

Self-Learning Thinking Machines for Security 

 

Good as that was, something has just emerged that leaves the above far in the evolutionary past. First developed under large government grants and utilized in the military and public sectors it is now being released for the private sector. It is remarkably inexpensive since it is simply added to any existing video surveillance system. Here the artificial intelligence has taken a huge leap. There is no longer any need to set rules for the system. It observes the environment for several weeks and learns what is normal behavior. From then on it reports any 

anomalies, that is anything that is abnormal behavior for what that camera sees. For example at a train station it is normal for people to jostle or bump into one another as they get onto a train. But at an ATM machine it is not normal for one person to bump into another. Back to the train station the shoving is normal but if someone ends up on the ground that is not normal. A fire starting or a cloud of smoke is not normal in most environments. Someone walking around a car a number of times in a paid parking lot is not normal. The artificial intelligence continues to classify events with great rapidity and increases its knowledge of what is normal behavior for that scene.

 

 This is called behavioral analytics. It means that existing surveillance cameras that have no intelligence and just record to a dvr, and have no ability to stop or prevent crimes from happening, can now be easily converted into this highly intelligent system. The artificial intelligence feels very much like what Ray Kurzweil wrote about. It is as if the future just became much closer. In practical terms it means that security has just taken an unimaginably large leap. The great bulk of the research and development cost was borne by the government and now the benefit is available to ordinary businesses and institutions for what is almost a negligible cost. It is extremely powerful artificial intelligence in the hands of companies and institutions faced with complex threats and security situations as well as ordinary trespass, theft and vandalism.

 

 

For further information contact Ken Hantman, President, Perimeter Protective Systems Inc.

866-633-3813, ken@perimeterpro.net or visit www.perimeterprotectivesystems.com