CPPMLBasics

See the trained network in action in your browser using WebGPU. Draw a digit and it will tell you what number it think you drew, in real time: https://electronicarts.github.io/gigi/Demos/mnist/index.html

The code repo to go with the blog posts and YouTube video:

1. Blog Part I: https://www.ea.com/seed/news/machine-learning-game-devs-part-1
2. Blog Part II: https://www.ea.com/seed/news/machine-learning-game-devs-part-2
3. Blog Part III: https://www.ea.com/seed/news/machine-learning-game-devs-part-3
4. All 3 parts in one video: https://www.youtube.com/watch?v=sTAqWRsEiy0

The Data folder contains the mnist hand drawn digits in their source format. Executing the Training project will extract those images into pngs.

The Demo folder contains the C++ DX12 demo which allows you to draw numbers, and have the neural network run on the GPU to try and classify what number you drew.

The Training folder contains the C++ code to train the neural network on the mnist data.

The Exercises folder contains the exercises that go along with the article.

Authors

Search for Extraordinary Experiences Division (SEED) - Electronic Arts
http://seed.ea.com
We are a cross-disciplinary team within EA Worldwide Studios.
Our mission is to explore, build and help define the future of interactive entertainment.

This code accompanies articles written by Alan Wolfe (blog.demofox.org)