public void Setup()
{
algorithm = new BSWTrainingAlgorithm();
var trainedModel = algorithm.Train(TestEntities.GetTrainingSamples());
bswNeuralNetwork = new BSWNeuralNetwork(trainedModel);
}
private void loadModelButton_Click(object sender, EventArgs e)
{
OpenFileDialog fileDialog = new OpenFileDialog();
fileDialog.Filter = "Json|*.json";
if (fileDialog.ShowDialog() != DialogResult.OK)
{
return;
}
initialTrainingSamples = Persistence.ReadTrainingSamplesFromFile(fileDialog.FileName);
additionalTrainingSamples = new List <Sample>();
bswTrainingAlgorithm = new BSWTrainingAlgorithm();
var initialModel = bswTrainingAlgorithm.Train(initialTrainingSamples);
bswNeuralNetwork = new BSWNeuralNetwork(initialModel);
}
private void InitializedBSW()
{
try
{
initialTrainingSamples = Persistence.ReadTrainingSamplesFromFile("SavedTrainingSamples\\initialTrainingSamples.json");
bswTrainingAlgorithm = new BSWTrainingAlgorithm();
var initialModel = bswTrainingAlgorithm.Train(initialTrainingSamples);
bswNeuralNetwork = new BSWNeuralNetwork(initialModel);
ShowBSWHiddenLayerNodes();
}
catch (Exception)
{
initialTrainingSamples = new List <Sample>();
}
additionalTrainingSamples = new List <Sample>();
}
private void TrainNetworkWith2GroupsOfNumbers(string class1, string class2)
{
var group1 = new List <int> {
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 50
};
var group2 = new List <int> {
60, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99
};
var group1Samples = group1.Select(x => new Sample {
InputVector = SerialCoding.GetSeriallyCodedValue(x, 100), OutputClass = class1
});
var group2Samples = group2.Select(x => new Sample {
InputVector = SerialCoding.GetSeriallyCodedValue(x, 100), OutputClass = class2
});
var allSamples = group1Samples.Concat(group2Samples).ToList();
var trainedModel = algorithm.Train(allSamples);
bswNeuralNetwork = new BSWNeuralNetwork(trainedModel);
}
private void TrainCurrentShape(ShapeType shapeType)
{
if (!strokes.Any())
{
return;
}
var trainingSample = Preprocessing.GetTrainingSampleFromShapeFeatures(currentShapeFeatures, shapeType);
additionalTrainingSamples.Add(trainingSample);
var allTrainingSamples = initialTrainingSamples.Concat(additionalTrainingSamples).ToList();
try
{
bswTrainingAlgorithm = new BSWTrainingAlgorithm();
var trainedModel = bswTrainingAlgorithm.Train(allTrainingSamples);
bswNeuralNetwork = new BSWNeuralNetwork(trainedModel);
}
catch (Exception e) { }
ShowBSWHiddenLayerNodes();
}
