Drugi jezik na kojem je dostupan ovaj članak: Bosnian
Source: ArsTechnica
Robots taking over the world! How many times have you heard such statements, and even feared them deep down? Well, there’s one robot whose fabulous fail (or win) gives us some assurances that bots are not poised to take over – yet.
A research scientist and neural network goofball Janelle Shane wondered about the origins of the ridiculous names of some colors, and took her wondering a step further. Shane decided to train a neural network to generate new paint colors, complete with appropriate names.
On her Tumblr account, Shane wrote: “For this experiment, I gave the neural network a list of about 7,700 Sherwin-Williams paint colors along with their RGB values. (RGB = red, green, and blue color values.) Could the neural network learn to invent new paint colors and give them attractive names?”
Shane told Ars that she chose a neural network algorithm called char-rnn, which predicts the next character in a sequence. So basically the algorithm was working on two tasks: coming up with sequences of letters to form color names, and coming up with sequences of numbers that map to an RGB value. As she checked in on the algorithm’s progress, she found that it was able to create colors long before it could actually name them reliably.
The longer it processed the dataset, the closer the algorithm got to making legit color names, though they are quite hilarious as you can see from the image.
When Shane cranked up “creativity” on the algorithm’s output, it gave her a violet color called “Dondarf” and a Kelly green called “Bylfgoam Glosd.” After churning through several more iterations of this process, Shane was able to get the algorithm to recognize some basic colors like red and gray, “though not reliably,” because she also gets a sky blue called “Gray Pubic” and a dark green called “Stoomy Brown.”
Judging from the results of this experiment, we can sleep tight, not worrying about robots swooping in – for now.
In case you want to see more Janelle Shane has unleashed her neural networks on everything from metal band names to Doctor Who episode titles. You can see all her work on her website.