/Neural-Network

(WIP) Neural network which can learn/identify handwritten numbers on a 28x28 pixel grid. The code is based on Michael Nielsen's code and data is from MNIST.

Primary LanguagePython

Neural-Network

(WIP) Modifying the code from Micheal Nielsen's neural network tutorial at http://neuralnetworksanddeeplearning.com/. The data is from MNIST and consists of 60k images. Currently at 95% accuracy. Plans:

  • Expand it to handwritten letters.
  • If the previous bullet works, then the ability to read any handwritten notes.
  • Try to optimize the code for better accuracy.
  • Apply the NN to text instead of images as they are easily available.