Neural network implementation, labeling numbers handwritten pictures from 0 to 9
The Neural network is composed of 4 parts. The input layer - 784 neurons corresponding to 784 pixels in the number PNG (28x28)
The first and second hidden layer - 128/64 neurons calculated by activating the sigmoid function on the weights * the pixels vector +bias
The output layer - 10 neurons corresponding to 10 tags (0-9) calculated by taking the second hidden layer output and activating the softMax function on it.
the most probable number guess from the neural network is taken as the chosen tag.
