| public RunEpoch ( double input, double output ) : double | ||
| input | double | Array of input vectors. |
| output | double | Array of output vectors. |
| 리턴 | double | |
public override ConfusionMatrix Execute()
{
//Create an network with one layer and one neuron in that layer
var network = new ActivationNetwork(new ThresholdFunction(), 3, 1);
//Bind the reference of the neuron
var neuron = network.Layers[0].Neurons[0] as ActivationNeuron;
//Create the Perceptron learning algorithm
//Library perceptron implements a single layer linear classifier
var teacher = new PerceptronLearning(network);
teacher.LearningRate = 0.1;
//Enrich the dimensions of the vectors, padding 1 to the end
var richTraining = AlgorithmHelpers.PaddDimension(trainingSet);
var richTesting = AlgorithmHelpers.PaddDimension(testSet);
//Training the network until the error is small enough
//or 500 hundred iterations have been computed
int epochs = 0;
while (true)
{
double error = teacher.RunEpoch(richTraining, trainingOutput);/// trainingSet.Length;
++epochs;
if (error < 0.025 * trainingSet.Length || epochs == 500) break;
}
var predicted = richTesting
.Select(x => neuron.Compute(x))
.Select(x => Convert.ToInt32(x))
.ToArray();
//Create a confusion matrix with the calculated parameters
ConfusionMatrix cmatrix = new ConfusionMatrix(predicted, expected, POSITIVE, NEGATIVE);
OnAlgorithmEnded(Enumerable.Repeat(neuron, 1), cmatrix);
return cmatrix;
}
// Worker thread
void SearchSolution( )
{
// prepare learning data
double[][] input = new double[samples][];
double[][] output = new double[samples][];
for ( int i = 0; i < samples; i++ )
{
input[i] = new double[variables];
output[i] = new double[1];
// copy input
for ( int j = 0; j < variables; j++ )
input[i][j] = data[i, j];
// copy output
output[i][0] = classes[i];
}
// create perceptron
ActivationNetwork network = new ActivationNetwork( new ThresholdFunction( ), variables, 1 );
ActivationNeuron neuron = network[0][0];
// create teacher
PerceptronLearning teacher = new PerceptronLearning( network );
// set learning rate
teacher.LearningRate = learningRate;
// iterations
int iteration = 1;
// statistic files
StreamWriter errorsFile = null;
StreamWriter weightsFile = null;
try
{
// check if we need to save statistics to files
if ( saveStatisticsToFiles )
{
// open files
errorsFile = File.CreateText( "errors.csv" );
weightsFile = File.CreateText( "weights.csv" );
}
// erros list
ArrayList errorsList = new ArrayList( );
// loop
while ( !needToStop )
{
// save current weights
if ( weightsFile != null )
{
for ( int i = 0; i < variables; i++ )
{
weightsFile.Write( neuron[i] + "," );
}
weightsFile.WriteLine( neuron.Threshold );
}
// run epoch of learning procedure
double error = teacher.RunEpoch( input, output );
errorsList.Add( error );
// show current iteration
SetText( iterationsBox, iteration.ToString( ) );
// save current error
if ( errorsFile != null )
{
errorsFile.WriteLine( error );
}
// show classifier in the case of 2 dimensional data
if ( ( neuron.InputsCount == 2 ) && ( neuron[1] != 0 ) )
{
double k = - neuron[0] / neuron[1];
double b = - neuron.Threshold / neuron[1];
double[,] classifier = new double[2, 2] {
{ chart.RangeX.Min, chart.RangeX.Min * k + b },
{ chart.RangeX.Max, chart.RangeX.Max * k + b }
};
// update chart
chart.UpdateDataSeries( "classifier", classifier );
}
// stop if no error
if ( error == 0 )
break;
iteration++;
}
// show perceptron's weights
ListViewItem item = null;
ClearList( weightsList );
for ( int i = 0; i < variables; i++ )
{
item = AddListItem( weightsList, string.Format( "Weight {0}", i + 1 ) );
AddListSubitem( item, neuron[i].ToString( "F6" ) );
}
item = AddListItem( weightsList, "Threshold" );
AddListSubitem( item, neuron.Threshold.ToString( "F6" ) );
// show error's dynamics
double[,] errors = new double[errorsList.Count, 2];
for ( int i = 0, n = errorsList.Count; i < n; i++ )
{
errors[i, 0] = i;
errors[i, 1] = (double) errorsList[i];
}
errorChart.RangeX = new DoubleRange( 0, errorsList.Count - 1 );
errorChart.RangeY = new DoubleRange( 0, samples );
errorChart.UpdateDataSeries( "error", errors );
}
catch ( IOException )
{
MessageBox.Show( "Failed writing file", "Error", MessageBoxButtons.OK, MessageBoxIcon.Error );
}
finally
{
// close files
if ( errorsFile != null )
errorsFile.Close( );
if ( weightsFile != null )
weightsFile.Close( );
}
// enable settings controls
EnableControls( true );
}
// Worker thread
void SearchSolution( )
{
// prepare learning data
double[][] input = new double[samples][];
double[][] output = new double[samples][];
for ( int i = 0; i < samples; i++ )
{
input[i] = new double[2];
output[i] = new double[classesCount];
// set input
input[i][0] = data[i, 0];
input[i][1] = data[i, 1];
// set output
output[i][classes[i]] = 1;
}
// create perceptron
ActivationNetwork network = new ActivationNetwork( new ThresholdFunction( ), 2, classesCount );
ActivationLayer layer = network[0];
// create teacher
PerceptronLearning teacher = new PerceptronLearning( network );
// set learning rate
teacher.LearningRate = learningRate;
// iterations
int iteration = 1;
// statistic files
StreamWriter errorsFile = null;
StreamWriter weightsFile = null;
try
{
// check if we need to save statistics to files
if ( saveStatisticsToFiles )
{
// open files
errorsFile = File.CreateText( "errors.csv" );
weightsFile = File.CreateText( "weights.csv" );
}
// erros list
ArrayList errorsList = new ArrayList( );
// loop
while ( !needToStop )
{
// save current weights
if ( weightsFile != null )
{
for ( int i = 0; i < classesCount; i++ )
{
weightsFile.Write( "neuron" + i + ";" );
weightsFile.Write( layer[i][0] + ";" );
weightsFile.Write( layer[i][1] + ";" );
weightsFile.WriteLine( layer[i].Threshold );
}
}
// run epoch of learning procedure
double error = teacher.RunEpoch( input, output );
errorsList.Add( error );
// save current error
if ( errorsFile != null )
{
errorsFile.WriteLine( error );
}
// show current iteration
iterationsBox.Text = iteration.ToString( );
// stop if no error
if ( error == 0 )
break;
// show classifiers
for ( int j = 0; j < classesCount; j++ )
{
double k = - layer[j][0] / layer[j][1];
double b = - layer[j].Threshold / layer[j][1];
double[,] classifier = new double[2, 2] {
{ chart.RangeX.Min, chart.RangeX.Min * k + b },
{ chart.RangeX.Max, chart.RangeX.Max * k + b }
};
// update chart
chart.UpdateDataSeries( string.Format( "classifier" + j ), classifier );
}
iteration++;
}
// show perceptron's weights
weightsList.Items.Clear( );
for ( int i = 0; i < classesCount; i++ )
{
string neuronName = string.Format( "Neuron {0}", i + 1 );
// weight 0
ListViewItem item = weightsList.Items.Add( neuronName );
item.SubItems.Add( "Weight 1" );
item.SubItems.Add( layer[i][0].ToString( "F6" ) );
// weight 1
item = weightsList.Items.Add( neuronName );
item.SubItems.Add( "Weight 2" );
item.SubItems.Add( layer[i][1].ToString( "F6" ) );
// threshold
item = weightsList.Items.Add( neuronName );
item.SubItems.Add( "Threshold" );
item.SubItems.Add( layer[i].Threshold.ToString( "F6" ) );
}
// show error's dynamics
double[,] errors = new double[errorsList.Count, 2];
for ( int i = 0, n = errorsList.Count; i < n; i++ )
{
errors[i, 0] = i;
errors[i, 1] = (double) errorsList[i];
}
errorChart.RangeX = new DoubleRange( 0, errorsList.Count - 1 );
errorChart.UpdateDataSeries( "error", errors );
}
catch ( IOException )
{
MessageBox.Show( "Failed writing file", "Error", MessageBoxButtons.OK, MessageBoxIcon.Error );
}
finally
{
// close files
if ( errorsFile != null )
errorsFile.Close( );
if ( weightsFile != null )
weightsFile.Close( );
}
// enable settings controls
EnableControls( true );
}
public void LearnDemo()
{
ActivationNetwork network = new ActivationNetwork(new ThresholdFunction(), 2, 1);//Constants.StoneCount
PerceptronLearning teacher = new PerceptronLearning(network);
double[][] input = new double[4][];
double[][] output = new double[4][];
input[0] = new double[] { 0, 0 };
output[0] = new double[] { 0 };
input[1] = new double[] { 0, 1 };
output[1] = new double[] { 0 };
input[2] = new double[] { 1, 0 };
output[2] = new double[] { 0 };
input[3] = new double[] { 1, 1 };
output[3] = new double[] { 1 };
double error = 1.0;
while (error > 0.001) {
error = teacher.RunEpoch(input, output);
}
var k = network.Compute(new double[] { 0.9, 0.7 });
var o = network.Output;
network.Save("a.txt");
}
//The Single Layer Perceptron classifier algorithm
//It uses an Activation Network library object, with one layer and one neuron to the layer
//To train the network it uses a PerceptronLearning library object
private ConfusionMatrix RunPerceptron(Double[][] trainingSet, Double[][] trainingOutput, Double[][] testSet, int[] expected)
{
//Create an network with one layer and one neuron in that layer
ActivationNetwork network = new ActivationNetwork(new ThresholdFunction(), 3, 1);
//Bind the reference of the neuron
ActivationNeuron neuron = network.Layers[0].Neurons[0] as ActivationNeuron;
//Create the Perceptron learning algorithm
//Library perceptron implements a single layer linear classifier
PerceptronLearning teacher = new PerceptronLearning(network);
teacher.LearningRate = 0.1;
//Enrich the dimensions of the vectors, padding 1 to the end
var richTraining = UtilityProvider.PaddDimension(trainingSet);
var richTesting = UtilityProvider.PaddDimension(testSet);
//Training the network until the error is small enough
//or 500 hundred iterations have been computed
int epochs = 0;
while (true)
{
double error = teacher.RunEpoch(richTraining, trainingOutput);/// trainingSet.Length;
++epochs;
if (error < 0.025 * trainingSet.Length || epochs == 500) break;
// Console.Write("Iter: " + epochs + " " + error + "\n");
}
var predicted = richTesting
.Select(x => neuron.Compute(x))
.Select(x => Convert.ToInt32(x))
.ToArray();
// Calculate the coordinates of the classifier
double k = (neuron.Weights[1] != 0) ? (-neuron.Weights[0] / neuron.Weights[1]) : 0;
double b = (neuron.Weights[1] != 0) ? (-((ActivationNeuron)neuron).Threshold / neuron.Weights[1]) : 0;
// Create the line and feed it to the data series
double[,] classifier = new double[2, 2]{
{ perceChart.RangeX.Min, perceChart.RangeX.Min * k + b},
{ perceChart.RangeX.Max, perceChart.RangeX.Max * k + b}
};
//For test, assume 1 as positive and 0 as negative
int positive = 0;
int negative = 1;
//Create a confusion matrix with the calculated parameters
ConfusionMatrix cmatrix = new ConfusionMatrix(predicted, expected, positive, negative);
//Update the most accurate classification
if (MostAccuratePerceptron == null || cmatrix.Accuracy > MostAccuratePerceptron.Item2.Accuracy)
{
MostAccuratePerceptron = Tuple.Create(classifier, cmatrix);
}
return cmatrix;
}