In this day and age, there are many things or people that we would like to be able to identify simply from their appearance. From judging whether someone has donned a disguise to more sinister implications, such identification would be critical for particular situations. A software company in Israel says it has developed a facial recognition software that they claim can identify terrorists and pedophiles, proving it by identifying…poker players.
According to Yahoo! Tech writer Dyllan Furness, The Israeli company Faception (which is a startup that is currently looking for funding) has developed this software and demonstrated it at an unnamed poker tournament. Yahoo! states that the software analyzed 50 amateur players that were in the tournament and, after comparing them to a database of poker professionals, picked four players that the software determined would excel in the event. Of those four players, two of them – amateurs, mind you – made the final table and would finish in the top three positions (it is not stated whether they won or not).
The theory that Faception’s software runs off is that facial profiling is supported by two genetic research observations, Furness says. DNA affects a person’s personality and also determines their facial features. Putting these two “genetic expressions” together and having a viable database to work from, the facial recognition software determines who has the potential to become whatever they company is searching for…be it poker pros, serial killers, geniuses or terrorists. The program has 15 different classifiers in its system and, according to its creators at Faception, the software is 80% accurate with its predictions.
There are some problems with the system that are pointed out, however. The determinations from the facial recognition software are only as good as the data that is put into the system (otherwise known as “garbage in, garbage out” or GIGO). The computer program is, after all, using an algorithm to qualify its decisions. There is an example of a computer program that correctly differentiated between wolves and dogs; it was quite proficient in its task, but the researchers found an error that led them to believe that the software wasn’t “differentiating” the two creatures – the computer program learned by identifying snow in the background of photos that it was more likely a wolf than a dog, which normally aren’t photographed in the snow.
The Washington Post’s Matt McFarland also looked into Faception’s work and potentially revealed further issues. “Can I predict that you’re an ax murderer by looking at your face and therefore I should arrest you?” said Pedro Domingos, a professor of computer science at the University of Washington and author of “The Master Algorithm.” “You can see how this would be controversial.” Domingos also pointed out that the software program may focus on one particular trait – such as a beard – and develop inaccurate predictions from such information.
The folks at Faception do not seem to be fazed by the naysayers, however. “We understand the human much better than other humans understand each other,” a Faception chief executive, Shai Gilboa, boasted to McFarland. “Our personality is determined by our DNA and reflected in our face. It’s a kind of signal.” They also state that they have a signed deal with an undisclosed security agency that will use the software to fight terrorism.
It might have been a joke previously – how all the great poker players were named Phil, for example, or they all were successful backgammon players – but what if there is something to the Faception theory and their software? Could we be identified by our faces, with them dictating what our future course may be from an examination of its contours? Or is it just another “snake oil” sales gimmick by a company to attempt to fool people?
