Thursday, November 19, 2015

Google and Facebook inverted to ‘sign’ their users with the facial recognition – Economíahoy.mx

Carlos Bueno – 21:40 – 11/17/2015
  • The technological treasure such a comprehensive knowledge users who destroys intimacy

  • Some systems are already able to recognize, even if we are back

The Big Brother is watching technology users from all loopholes. Scare discover the possibilities available to Google or Facebook if to perform a thorough profile of each user. The search engine, for example, has access to mails from Gmail, in addition to web browsing Android, consulting its mobile applications store, comments on social networks, historical or shopping trips by plane, plus downloads or trips by car or foot. Facebook, WhatsApp also owns, could do the same with the likes and dislikes of each Internet.

The next goal will be the identification and geolocation through the face. To the thousands of cameras that track the movements of every citizen also binds the innocence of the Internet. And the habit of labeled at the photos that are shared on social networks does not help preserve that privacy.

The face recognition has become a priority for commercial companies like Facebook and Google on the one hand, and police and security forces on the other. The former allowed to multiply the huge amount of data and information that already manage users and potential consumers. In the second, they would greatly facilitate things in detecting potential threats, locate fugitives, reduce the level of alarm … But fate is fickle and, in this case, nature wants to put it very difficult to technology.

No two are alike

The fact that no two identical faces in the world is greatly complicating the work of the engineers to finish their facial recognition systems. Large Internet companies have for years invested millions to give to the system identify anyone by a simple image sums. The objective is to emulate the human brain. And they do not go misguided, because they have calculated that people recognize someone with a 97.53 percent effective. Well, those test systems have already achieved an effectiveness of 97.25 percent. To do this, as have those responsible, they have had to make an explicit 3D modeling in nine layers, making up a whopping 120 million more parameters.

In the case of Facebook, explain that they have trained this model using the facial larger set of data obtained to date. Obviously, if we consider that, true to its name of faces–book each user get a few portraits from which to choose for comparison. So they pulled file and have seized the photos that hang on the main social network become the property of them (or did not read the terms before opening the profile). For starters, it has been sufficient to four million labeled images belonging to 4,000 people. Thus, each individual recognize handled thousand images to compare different features.

“We do not want to imagine what it would mean for Facebook access that information, to know what we do and where we are at all times, conduct that our real life tracking to detect behaviors, social habits, real interests … “forward Javier Sirvent, technlogy evangelist. But the team of artificial intelligence Facebook, led by Yann LeCun not want to plant alarms. “These companies act giving small steps and, depending on the reaction of the people, are one way or another,” recalls Sirvent. In this sense, Facebook and activated his service automatic labeling of images, but decided to withdraw a few weeks for the confusion it caused.

According LeCun, “we will not see this new technology in use until people are ready to accept our reality.” And it even can recognize seeing you back. In an interview with Yahoo, alleging that it is a “scientific experiment”, something that makes us stay much calmer. “We publish everything we do to be seen by the research community. We’re trying to advance science,” he added.

At the recent meeting, artificial intelligence chief admitted that Facebook are already able to recognize someone even face half covered with hair , a hat … To reach that point, they have also studied your complexion, her clothes, her style … “facial recognition products have been launched in most countries, but not in places where people are uncomfortable with that application” He explained. “We do not want people angry and, for this reason, we have chosen not to implement facial recognition in Europe, where users do not feel good about this technology,” he added.



Facebook versus FBI

This large database can talk face recognition system Facebook and even better than it can handle the FBI. It would not be surprising if we consider that handles Facebook file is 1,500 million people worldwide. Google is also not far behind, especially since your last launched application for Android, Google Photos. Simply upload this cloud all images and will appear as listed by them and even emotions.

If we accept that Google resize the image, change give us unlimited space for that cloud. “These companies work like this: First you get a search engine, then a mail account, then the maps later this cloud … now, and continue to manage change as much information as possible about each user” recalls Sirvent, who acknowledges having risen to 30,000 images Google Photos. The reason? “For the simple reason that in the end makes up for what you get what you ask for you”, is justified.



Other news
sponsored content
LikeTweet

No comments:

Post a Comment