My Attempt at a supervised neural network, it can be reformatted easily for a variety of applications and i plan to make a an unsupervised varient for estimating price change in crypto-currency, i plan to upload that code when more progress has been made

Ideas and inspiration were drawn from the following litterature: "Hands On with Machine Learning with Scikit-Learn and TensorFlow" - Aurélien Géron "Build your own neural Network" - Tariq Rashid

Developed using Jupyter notebooks, Matplotlib, Scikit and Numpy.

the Training Data in this instance is 60,000 points of data the Test uses only 10,000

Currently I have refined it to 96% accuracy

Trainining and Test Data Can be found here:

Test Data: https://pjreddie.com/media/files/mnist_test.csv Training Data: https://pjreddie.com/media/files/mnist_train.csv