Google X, the supercomputer and the process of automatic learning

Straight from the secret room, or almost, laboratory Gooogle X comes the new advanced project signed big G that has as ultimate goal to study and better understand the learning ability of computer networks.

At the heart of the project is, specifically, a neural simulator which precisely studying the intelligence of machines.

The network has been built by researchers from the Google team is is based on 1000 computers networked to a total of 16 000 different computational units so as to simulate a part of the visual cortex of the human brain.

After setting the system Google has then fed to the network which have created YouTube videos starring cats.

The final result was that the network, thus, is able to recognize one the whole concept of cat after examining about 10 million digital images of happy home inside the various videos.

With this experiment and its success Google has highlighted a key feature of learning algorithms: they are able to achieve high performance if they have access to a huge amount of data.

Taking this into account and considering the great success of the experiment Google has now decided to move the project from the laboratory to the X Division engaged on research and various web and any related services with the aim to develop learning algorithms that allow computer enhance, not a little, the search methodology between large amounts of information on a network such as, for example, in the case of images, or even relative to the voice recognition.

Posted in: automatic learning, google x, lab x, news, simulate human brain, supercomputer

Reazioni: