Minimalistic neural network implementation with purpose for classification of handwritten crosses on spreadsheets.
Project started for experimenting for a final's project for a student course I supervised. Remained hidden to prevent cheating, later abandoned and moved to private GitLab project due to preferences of coworkers.
For convenience data is included in project. Here we only have the core minimalized network to train which already does the job and everyone should be able to implement. Based on the amazing description of MNIST by Michael Nielsen with related repo
This project is dead, in case you are interested in a full implementation feels free to ask me or implement yourself using the detailed description.