Useful Links to Color Lists about color-names HOT 3 OPEN

Wow @tajmone thanks for this!

• I think I had added tthe colors for "the color thesaurus" manually, I will check if I actually did or only planed to ;)
• I had seen the colors from http://aiweirdness.com/post/160776374467/new-paint-colors-invented-by-neural-network, and I also found it super interesting, but too many of the names are actually not that good, I think it could be interesting to it with our list to check out, if the outcome would be better. I did not dig into machine learning at all for now though

from color-names.

I had seen the colors from http://aiweirdness.com/post/160776374467/new-paint-colors-invented-by-neural-network, and I also found it super interesting, but too many of the names are actually not that good,

Yes, I had also come to the same conclusion, but I was quite impressed by the whole approach and many colors did proof to be interesting. Hard to know if IA could contribute in this direction, so right now we might take it with a pinch of salt and consider it just a curiosity — but it's definitely an experiment worth mentioning in any colors-naming project.

I think it could be interesting to it with our list to check out, if the outcome would be better. I did not dig into machine learning at all for now though.

It might be doable, and there are some good neural network frameworks and libraries in the Node/JS echo system that could be used:

• https://brain.js.org/ (this JS demo deals with color contrast reckognition)
• http://caza.la/synaptic/#/

JS tutorials on the topic are plenty:

• Neural networks in JavaScript
• How to create a Neural Network in JavaScript in only 30 lines of code

Setting up a simple perceptron NNA should be straightfoward, and it might be used to develop an alogrithm to learn from the manual decision on how duplicate color entries [are being/should be] handled. Here's an implementation in 140 lines:

• https://gist.github.com/primaryobjects/dfb8927f9f0ca21b6a24647168cead41

There are enough entries in the colors lists to provide and adequate amount of starting data, the trickiest part would be to establish the criteria by which the machine should classify the problem and learn its lessons.

Could become an "experimental" feature on the side.

from color-names.

Would love to dive into that! Definitely let us know here, when you got some time!

from color-names.