The one-size-fits-all security software installed by IT departments is often not the best possible solution as it can produce false positives as well as miss actual attacks. Researchers at Intel Research Berkeley have recognized that problem and are working on laptop-based security software that adjusts to the way an individual uses the internet to better detect malicious activity.

“One reason security breaches are so rampant is that most of our machines look the same,” says Nina Taft, a researcher with the program. “When a hacker breaks into one machine, he can break into all of them… We’re trying to inject diversity into computers.”

Traditional security software has a preset threshold. When internet activity goes above that level, the software triggers an alarm suggesting that the computer might be infected. These kinds of infections are mainly due to botnets, which are enormous quantities of infected computers acting together to send out spam and do other malicious deeds. However, users who use the internet more than average could have to deal with frequent false alarms, and users who barely use their connection might never know if their computer got infected. Read more »

Researchers at the Department of Artificial Intelligence (DIA) of Universidad Politécnica de Madrid have developed an algorithm that is capable of detecting a person’s facial expressions. The software can analyze 30 images per second and categorize expressions as anger, disgust, fear, happiness, sadness, or surprise.

The software is capable of following a face around and spot its emotions anywhere in a frame. It can even work on recorded video, recognizing what expressions a face in the video is showing.

Although only a prototype, the software is capable of working on a regular desktop or laptop computer, allowing you to take it with you and use a webcam to analyze your expressions (in case you don’t know how you’re feeling at any given time).

This software has great potential for a wide variety of uses. For example, it might eventually be used as a sort of detector to see if a person is lying.

Check out a cool video of the algorithm in action here.

Psychology researchers at the Glasgow University have been able to eliminate the variable effects of age, hairstyle, expression, lighting, and different camera equipment in a face recognition algorithm, resulting in a “100% success rate“. They achieved this by producing a composite “average face” for a person from 20 different pictures.

From the two researchers:
“We modeled human familiarity by using image averaging to derive stable face representations from naturally varying photographs. This simple procedure increased the accuracy of an industry standard face-recognition algorithm from 54 per cent to 100 per cent, bringing the robust performance of a familiar human to an automated system.”

Of course, there’s a catch. By 100% the researchers really mean that the detection in each one of their tests was correct, not that they are sure it would work 100% of the time in real life.

Although probably not perfect, this is still a valuable increase from the old detection methods and could help in today’s facial recognition applications. Of course, this is provided that 20 photos of the person already exist in some database. An image of John Travolta’s “average” face after the jump. Read more »