Open luiscarbonell opened 5 years ago
Make a simple NN which learns to play soccer. Always good to impress ppl :)
@Pummelchen @christianechevarria
Something like:
???
Maybe with OpenAI's HTTP API version of their Gym: https://github.com/openai/gym-http-api
Flappy bird demo's been the standard when showing off neuroevolution.
@dan-ryan Will put it as a placeholder here, for now: https://github.com/xviniette/FlappyLearning
No, something simples like this:
https://www.youtube.com/watch?v=uTzdif1AUlw
Can be even simpler by just one player moving and the other waits till its his turn. But the NN needs to learn what to do with the ball and that it should go to the opposite goal.
Ages back I suggested doing a Mario example with Neataptic. It would really show the power of Carrot's neuroevolution. https://github.com/wagenaartje/neataptic/issues/21
We could also port the Neataptic text prediction example (also listed in the above issue): https://jsfiddle.net/wagenaartje/8r4mwe3d/5/
From reinforce.js, here's an example: https://mvrahden.github.io/learning-agents/
Also ran into this: https://github.com/naptha/tesseract.js
Could make for some interesting demos
@dan-ryan Thank you for the Snake - just posting it here.
NOTE: Another snake demo.
These games are amazing for demo!!!
Other Games:
Google Dinosaur Loading Screen:
Hey everyone! Implemented a flappy-bird demo and playing around with it lately:
https://liquidcarrot.io/example.flappy-bird/
Thinking about making this a learning sandbox for new users with code examples, explanations on GAs, and generally showing off features of the library
Here's a screenshot:
Examples & Use Cases
It would be great if we could get a strong set of use cases and examples built out for different communities to be able to use in their individual projects - and maybe add them to the README.md, @christianechevarria.
As a first contender, @mateogianolio created this great - really simple - OCR example with Synaptic.js.
I think this could be a really cool visual - maybe implement it as a GAN and have the generative NN do a live display.