.NET libarary used to implement and train neural networks using C#
For a full demo, including how to package MNIST data for the network, this repo shows usage of the library: https://github.com/JEMeyer/MNIST-Neural.NET
To see how to get started with the library, head over to the wiki. There are two articles, one if you are using the NuGet package, and one for if you are pulling down the repository and building locally.
My goal with this project is to allow neural networks to be more approachable. A lot of the 'matching up' and intricate details that might be annoying to figure out have all been abstracted away. If you have watched a simple tutorial on neural networks and understand how they work at a high level, you should be able to use this libaray. That is my goal, to allow any developer to be able to build their own projects using a neural network with only a basic understanding of how they work. More features will be added with time, so the NuGet package and repo might be out of sync, which is why I have split up the getting started depending on what codebase you are using.
If you have any questions, feel free to open an issue on GitHub.
Thanks!