Handwritten Recognition

This project will be my first real Artificial Intelligence (AI). A simple artificial neural network from scratch. I'll dissect it and learn how AI works. (*.idx1-ubyte and *.idx3-ubyte)

Prerequisites Files:

1. Handwritten.Recognition.py

• This is the main source code of the project.

2. train-images.idx3-ubyte

• The MNIST database of handwritten digits, has a training set of 60,000 examples.

  Source: http://yann.lecun.com/exdb/mnist/

3. train-labels.idx1-ubyte

• This database file is the result of handwritten digits, for training of the neural network.

  Source: http://yann.lecun.com/exdb/mnist/

  python -m pip install --user numpy scipy matplotlib ipython jupyter pandas sympy nose

Phasing of the Project:

1. Downloading Handwritten Digits Datasets. (Done.)
2. Reading the Datasets. (Done)
3. This Project will be doing a Feedforward Neural Network as an artificial neural network to read the handwritten digits.

• • 3a. Backpropagation (Ongoing: Learning the Mathematics in Feedforward Neural Network)
• • 3b. Testing the with the datasets using Feedforward Neural Network.

4. Find more datasets to test the with code.
5. User input and the project will automatically detect the digits. (I cannot decide yet how the users will input.)

Next phase will be doing Backpropagation for the datasets I have collected.