A Google research team has trained a network of 1,000 computers wired up like a brain to recognise cats.
The team built a neural network, which mimics the working of a biological brain, that worked out how to spot pictures of cats in just three days.
The cat-spotting computer was created as part of a larger project to investigate machine learning.
lucy24
12:31 am on Jun 27, 2012 (gmt 0)
:: detour to read article ::
After only three days of work (equivalent to something like eight trillion years of human cogitation?), the computer can point to a picture and say "It's a cat!"
Now, if it could point to a picture and say "That's Matt Cutts's cat!" I'd be impressed.
engine
12:03 pm on Jun 27, 2012 (gmt 0)
hehehe
I actually thought there were other systems out there that were more impressive, but then, I might have seen it in a movie. ;)
jmccormac
12:43 pm on Jun 27, 2012 (gmt 0)
Now if they could develop an algorithm that recognised good websites... :)
Regards...jmcc
londrum
1:33 pm on Jun 27, 2012 (gmt 0)
thats probably very difficult to do, given all the different shapes, sizes and colours that cats come in. but you never know how they are doing it. if they have just spent 3 days storing up a bazillion million different pictures of cats, and compared each new one to them, then that's not really "thinking". it still hasnt got a clue what a cat is. its just found a picture that looks like what its looking at, and tagged it with 'cat'
what if you molded a cat out of a piece of stone. or carved one out of wood. would it still be able to recognise it as being a cat?