private void ResetNetwork()
{
network = new Network(io.Inputs, io.Outputs, (int)numNeurons.Value,
(int)numLayers.Value);
trainer = new Trainer(network, io);
trainer.LearningRate = (double)numRate.Value;
totalIterations = 0;
lblTrainingProgress.Text = "";
lineGraph.SetData(null);
lineGraph.Refresh();
}
static void Main(string[] args)
{
/*
* XOR problem.
* I seemed to have issues with only 2 neurons in the hidden layer.
* It never seemed to converge well, with the error calculation on the output
* oscillating between positive and negative.
*/
Network network = new Network(0.25, 1.0, 0, 2, 4, 1);
network.AddPattern(new Pattern { Input = { 0, 0 }, Output = { 0 } });
network.AddPattern(new Pattern { Input = { 0, 1 }, Output = { 1 } });
network.AddPattern(new Pattern { Input = { 1, 0 }, Output = { 1 } });
network.AddPattern(new Pattern { Input = { 1, 1 }, Output = { 0 } });
network.Cycle(10000);
network.PrintOutput();
/*
* Recognize the digits 0 - 9
* Represent digits with 4 x 5 grid, making input layer 20 neurons large.
* Output layer has 10 neurons, one for each digit.
* Again, this experiment is not entirely successful.
* The setup below only recognized 0, 1, 2, 5, 6, 7
*/
//Network network = new Network(0.1, 1.0, 0, 20, 40, 10);
//network.AddPattern(new Pattern
//{
// Input = { 1, 1, 1, 1,
// 1, 0, 0, 1,
// 1, 0, 0, 1,
// 1, 0, 0, 1,
// 1, 1, 1, 1 },
// Output = { 1, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
//});
//network.AddPattern(new Pattern
//{
// Input = { 0, 0, 0, 1,
// 0, 0, 0, 1,
// 0, 0, 0, 1,
// 0, 0, 0, 1,
// 0, 0, 0, 1 },
// Output = { 0, 1, 0, 0, 0, 0, 0, 0, 0, 0 }
//});
//network.AddPattern(new Pattern
//{
// Input = { 1, 1, 1, 1,
// 0, 0, 0, 1,
// 1, 1, 1, 1,
// 1, 0, 0, 0,
// 1, 1, 1, 1 },
// Output = { 0, 0, 1, 0, 0, 0, 0, 0, 0, 0 }
//});
//network.AddPattern(new Pattern
//{
// Input = { 1, 1, 1, 1,
// 0, 0, 0, 1,
// 1, 1, 1, 1,
// 0, 0, 0, 1,
// 1, 1, 1, 1 },
// Output = { 0, 0, 0, 1, 0, 0, 0, 0, 0, 0 }
//});
//network.AddPattern(new Pattern
//{
// Input = { 1, 0, 0, 1,
// 1, 0, 0, 1,
// 1, 1, 1, 1,
// 0, 0, 0, 1,
// 0, 0, 0, 1 },
// Output = { 0, 0, 0, 0, 1, 0, 0, 0, 0, 0 }
//});
//network.AddPattern(new Pattern
//{
// Input = { 1, 1, 1, 1,
// 1, 0, 0, 0,
// 1, 1, 1, 1,
// 0, 0, 0, 1,
// 1, 1, 1, 1 },
// Output = { 0, 0, 0, 0, 0, 1, 0, 0, 0, 0 }
//});
//network.AddPattern(new Pattern
//{
// Input = { 1, 1, 1, 1,
// 1, 0, 0, 0,
// 1, 1, 1, 1,
// 1, 0, 0, 1,
// 1, 1, 1, 1 },
// Output = { 0, 0, 0, 0, 0, 0, 1, 0, 0, 0 }
//});
//network.AddPattern(new Pattern
//{
// Input = { 1, 1, 1, 1,
// 0, 0, 0, 1,
// 0, 0, 0, 1,
// 0, 0, 0, 1,
// 0, 0, 0, 1 },
// Output = { 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }
//});
//network.AddPattern(new Pattern
//{
// Input = { 1, 1, 1, 1,
// 1, 0, 0, 1,
// 1, 1, 1, 1,
// 1, 0, 0, 1,
// 1, 1, 1, 1 },
// Output = { 0, 0, 0, 0, 0, 0, 0, 0, 1, 0 }
//});
//network.AddPattern(new Pattern
//{
// Input = { 1, 1, 1, 1,
// 1, 0, 0, 1,
// 1, 1, 1, 1,
// 0, 0, 0, 1,
// 0, 0, 0, 1 },
// Output = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 1 }
//});
//network.Cycle(10000);
//network.PrintOutput();
Console.In.ReadLine();
}
/// <summary>
/// Creates a new instance of trainer with the network to train and
/// the input/output to train for.
/// </summary>
/// <param name="network">The network to train</param>
/// <param name="io">The input/output to use</param>
public Trainer(Network network, TrainerIO io)
{
this.network = network;
this.io = io;
learningRate = io.LearningRate;
}