Friday, May 19, 2017

Meanwhile, In the Mind of AI

Researcher and lover of neural networks, Janelle Shane, was thinking about the strange and usual paint names and wondered what a neural network would name different paint shades. So she fed in 7,500 Sherwin Williams paint names and RGB values, and let the network go to work. The results is not only an interesting insight into neural networks and a higher-level view of how they work; it's pretty hilarious what the network came up with:
One way I have of checking on the neural network’s progress during training is to ask it to produce some output using the lowest-creativity setting. Then the neural network plays it safe, and we can get an idea of what it has learned for sure.

By the first checkpoint, the neural network has learned to produce valid RGB values - these are colors, all right, and you could technically paint your walls with them. It’s a little farther behind the curve on the names, although it does seem to be attempting a combination of the colors brown, blue, and gray.
Later in the training process, the neural network is about as well-trained as it’s going to be (perhaps with different parameters, it could have done a bit better - a lot of neural network training involves choosing the right training parameters). By this point, it’s able to figure out some of the basic colors, like white, red, and grey, although not reliably.

In fact, looking at the neural network’s output as a whole, it is evident that:
  1. The neural network really likes brown, beige, and grey.
  2. The neural network has really really bad ideas for paint names.

Stoner Blue is pretty nice. Think I'll skip Sindis Poop.

No comments:

Post a Comment