Hello Wilfred,
LL>> In 2012, a Stanford college professor (Andrew Ng) put together a team
LL>> of students who built a deep-learning system that analyzed 10 million
LL>> YouTube video stills and, without any human input at all, created a
LL>> correct understanding of a cat.
WvV> I "studied" hardly a dozen cats in my live time, and I probably
understand
WvV> them just as well... As far as you can understand cats. ;)
How do you know what you think you saw was a cat? I mean,
just because something has pointy ears, fur, and a tail does
not necessarily mean that it is a cat. Although it could be.
Generative Adversarial Networks (GANs) - A photo of a cat
is introduced. This is the definition of a cat. One network
generates images of cats that it shows to another network.
That other network's job is to determine whether the first
is showing a cat or something that looks like a cat. With
time, the analysis gets better and better.
That can tell you what a cat is. But can anyone, or anything,
truly understand cats?
--Lee
--
Erections, That's Our Game
---
This email has been checked for viruses by AVG.
http://www.avg.com
--- MesNews/1.08.05.00-gb
* Origin: news://eljaco.se (2:203/2)
|