Adsense Menu

Search Engine

Search Result

Monday, August 6, 2012

Google’s Artificial Brain Which Can Recognize A Cat

Google X laboratory has found some pretty cool: a refrigerator that can order your groceries when food runs low, an elevator that could reach the outer space, self-driving car. So it's no surprise that their most recent design is a function, the most advanced highest, the most amazing invention ever ... computer that cats love to watch YouTube?

Okay, it's a bit more sophisticated than that. Several years ago, scientists began creating Google neural networks for machine learning. Google X technique used for this project is called "learning in," a method that is defined by a very large scale. In layman's terms, they are connected 16 000 computer processors, and let them create a network to roam freely on the Internet so as to simulate the human brain learning.

Stanford University computer scientist Andrew Y. Ng, Google led the team in the feeding neural network of 10 million digital images of random YouTube videos. The machine was not "controlled," that is not told what or what features a cat a cat has, but just look at the data at random to be fed to it. Ng found that there is a small part of the "brains" of computers that teach themselves to recognize the cat. "It's basically invented the concept of a cat," said Google fellow Jeff Dean New York Times.

So Google may have created a machine that can teach itself. But what Ng and his team have done is not as new as you may think. Over the years, as the scale of the simulation software has been developed, machine learning systems have been developed; years ago, Microsoft scientists suggest that "learning in" techniques can be used to build computer systems to understand human speech. Google this computer is the cream of the crop X-twice as accurate as the other machines before. However, "it should be noted that our network is still small compared to the human visual cortex," the researchers wrote, "that a million times bigger in terms of the number of neurons and synapses."

After the "look" random pictures of random YouTube videos, neural networks create digital images of cats based on "memory" of the form that saw him in the picture. Cats do not make computers any particular cat, but imagine what a computer to be a cat. Plato has the form, and now Google has a computer generated picture of his cat.

Technology Review RSS Feeds

Blog Top Sites