A minimalistic neural network for educational purposes.

This project is a side product of a term paper on neural networks that I wrote in my first year on vocational school. The original paper discussed the general principles of neural networks and their advantages and disadvantages regarding modern (web based) applications.

The published code demonstrates these general priciples by creating a minimal neural network, which can be trained to recognize hand-written digits. Both training and testing are contolled by a simple graphical interface, which also provides some nice feedback on the current progress and success rates to the user.

All of the code is written in Java, so you need the Java Runtime Environment to run it. To compile it from source, the Java Development Kit is needed aswell.

As this is a finished project, contributing is not possible. Consider this repository as read-only. Of course, you should feel free to fork this repository as desired.

This project is licensed under the MIT License - see the LICENSE.md file for details.