The person whom you see in the image above is Stanford computer scientist Andrew Ng, who is seated beside an image of a cat which a neural network actually taught itself to identify, now how about that? The thing is, the amount of computing power required to achieve a seemingly simple exercise is nothing to sneeze at – we are talking about 16,000 computers after all. Google’s highly secretive X laboratory that is famous for previous inventions such as a self-driving car and augmented reality glasses, also housed a relatively tight knit group of researchers who started working on a human brain simulation for the past several years already.
When presented with 10 million digital images which were harvested from YouTube videos, this ‘brain’ from Google started to perform a search for cats, which is not too uncommon among humans either. The Internet, after all, is full of cat videos, but the results of this simulation surprised the team as it performed in a far superior manner compared to other efforts, where the accuracy level has doubled in terms of recognizing objects out of 20,000 distinct items listed. Just how will this brain simulator change the way the world works for the better? Only time can – and will, tell.
I cannot help but think of the expression as to how there is more than one way to skin a cat. In this case, there are 16,000 ways to recognize a cat – and each of them is a computer.
IBM Watson Smartphone Apps Coming Soon
Chrome App Launcher Beta For Mac OS X Released
Alleged Wireless Card For New Macs Hints Next-Gen Wi-Fi Support
Lenovo ThinkPad S3, S5 Leaks Show Aluminum-Cased Ultrabooks