This project use backpropagation to train a network for character recognition. Structure of the Network: 80 neurons in the Input layer, 100 neurons in the Hidden layer 10 neuron in the Output layer By default it uses a bias neuron in the Input and Hidden layers but, setting the Bias slider variable to zero, you can run the network with no biases. With the slider variable 'Font' you may choose the type of characters to recognize: the first two are fonts for digits and the last one is for letters (the first 10 of the alphabet). Meaning of values in slider variable Start: 0 - use random weights 1 - use trained weights 2 - use previous weights Slider variable 'eta' is the learning rate: more high the faster the algorithm runs but to fine tune, use a low number. The % of success is calculated on the last 100 attempts. Thanks to PutneyCat and his Basic Neural Network for the idea. Reference documentation: https://mattmazur.com/2015/03/17/a-step-by-step-backpropagation-example/
