Catalog · How the catalog works
neural-network-from-scratch
Neural network implementation from scratch for handwriting recognition based on the MNIST dataset.
- ★ 2 on GitHub
Anyone may press Interesting, signed in or not. It helps others find this project.
README
Neural Network from Scratch
This is a small project about neural networks and backpropagation that I did in school. It uses the MNIST database to learn handwriting recognition. The whole neural network was built from scratch in Delphi 10.2 without using any external libraries.
The implementation itself is pretty basic and lacks a lot of features you would want from a neural network, like automatic calculation of the validation accuracy and so on. But all in all, it works, and considering it was just an educational project, it was pretty cool, even though it was written in one of the most inappropriate programming languages for this use case.
[](https://github.com/DevTobias/neural-network-from-scratch/blob/HEAD/[https:/github.com/tobias-kaerst-software](https:/www.youtube.com/playlist?list=PLZHQObOWTQDNU6R1_67000Dx_ZCJB-3pi))The theoretical underpinnings were taken from the phenomenal Deep Learning YouTube series by 3Blue1Brown. If you are considering doing something like this yourself, these videos are a must.