The program will import network parameters from the file params.txt. The first line of the file must be the topology of network, and the second must be the eta, then alpha values seperated by a space.
Parameters can be changed during training. E and W, S and A keys will increase and decrease eta and alpha values respectively.
The network imports training data from training_data.txt. A python file is included that will generate various different data sets, but as long as it follows the correct format you can use your own data.
Use the g key to toggle between displaying neuron activation values and gradients.
Use the up and down arrow keys to increase or decrease timescale.
Bias neurons are indicated in blue, connection weights and neuron activation values are displayed on a blue-red-green spectrum (close to -1 is blue, close to 0 is red, close to 1 is green).
Displays number of trials, average mean squared error over last 1000 trials, eta, and alpha values.