This is a Rust implementation of a simple neural network for recognizing handwritten digits. It is trained on the MNIST dataset.
This is a self-contained project without any dependencies besides the Rust standard library. It was built as a learning exercise, without tremendous thought for efficiency. In particular, the number of memory allocations could be cut down significantly to increase performance.
Much thanks to Michael Nielsen for his book Neural Networks and Deep Learning. This project is primarily an implementation of the concepts and examples taught there.