static void Main(string[] args)
{
Console.WriteLine("Neural Network Demo using .NET by Sebastian Brandes");
Console.WriteLine("Data Set: Breast Cancer Wisconsin (Diagnostic), November 1995");
// Source: http://archive.ics.uci.edu/ml/machine-learning-databases/breast-cancer-wisconsin/
Console.WriteLine();
#region Data Generation
Console.WriteLine("Loading source file and generating data sets...");
var rows = File.ReadAllLines(sourceFile);
var data = new List<double[]>();
foreach (var row in rows)
{
var values = row.Split(',');
var observation = new double[values.Length];
for (int i = 0; i < values.Length; i++)
{
double.TryParse(values[i], out observation[i]);
}
data.Add(observation);
}
List<double[]> trainData;
List<double[]> testData;
Helpers.GenerateDataSets(data, out trainData, out testData, 0.8);
Console.WriteLine("Done!");
Console.WriteLine();
#endregion
#region Normalization
Console.WriteLine("Normalizing data...");
List<double[]> normalizedTrainData = Helpers.NormalizeData(trainData, inputColumns);
List<double[]> normalizedTestData = Helpers.NormalizeData(testData, inputColumns);
Console.WriteLine("Done!");
Console.WriteLine();
#endregion
#region Initializing the Neural Network
Console.WriteLine("Creating a new {0}-input, {1}-hidden, {2}-output neural network...", numInput, numHidden, numOutput);
var nn = new NeuralNetwork(numInput, numHidden, numOutput);
Console.WriteLine("Initializing weights and bias to small random values...");
nn.InitializeWeights();
Console.WriteLine("Done!");
Console.WriteLine();
#endregion
#region Training
Console.WriteLine("Beginning training using incremental back-propagation...");
nn.Train(normalizedTrainData.ToArray(), maxEpochs, learnRate, momentum, weightDecay);
Console.WriteLine("Done!");
Console.WriteLine();
#endregion
#region Results
double[] weights = nn.GetWeights();
Console.ForegroundColor = ConsoleColor.Green;
Console.WriteLine("Final neural network weights and bias values:");
Console.ResetColor();
Helpers.ShowVector(weights, 10, 3, true);
Console.WriteLine();
double trainAcc = nn.Accuracy(normalizedTrainData.ToArray());
Console.WriteLine("Accuracy on training data = " + trainAcc.ToString("F4"));
double testAcc = nn.Accuracy(normalizedTestData.ToArray());
Console.WriteLine("Accuracy on test data = " + testAcc.ToString("F4"));
Console.WriteLine();
//Console.ForegroundColor = ConsoleColor.Green;
//Console.WriteLine("Raw results:");
//Console.ResetColor();
//Console.WriteLine(nn.ToString());
#endregion
Console.ReadKey();
}
static void Main(string[] args)
{
Console.WriteLine("Neural Network Demo using .NET by Sebastian Brandes");
Console.WriteLine("Data Set: Breast Cancer Wisconsin (Diagnostic), November 1995");
// Source: http://archive.ics.uci.edu/ml/machine-learning-databases/breast-cancer-wisconsin/
Console.WriteLine();
#region Data Generation
Console.WriteLine("Loading source file and generating data sets...");
var rows = File.ReadAllLines(sourceFile);
var data = new List <double[]>();
foreach (var row in rows)
{
var values = row.Split(',');
var observation = new double[values.Length];
for (int i = 0; i < values.Length; i++)
{
double.TryParse(values[i], out observation[i]);
}
data.Add(observation);
}
List <double[]> trainData;
List <double[]> testData;
Helpers.GenerateDataSets(data, out trainData, out testData, 0.8);
Console.WriteLine("Done!");
Console.WriteLine();
#endregion
#region Normalization
Console.WriteLine("Normalizing data...");
List <double[]> normalizedTrainData = Helpers.NormalizeData(trainData, inputColumns);
List <double[]> normalizedTestData = Helpers.NormalizeData(testData, inputColumns);
Console.WriteLine("Done!");
Console.WriteLine();
#endregion
#region Initializing the Neural Network
Console.WriteLine("Creating a new {0}-input, {1}-hidden, {2}-output neural network...", numInput, numHidden, numOutput);
var nn = new NeuralNetwork(numInput, numHidden, numOutput);
Console.WriteLine("Initializing weights and bias to small random values...");
nn.InitializeWeights();
Console.WriteLine("Done!");
Console.WriteLine();
#endregion
#region Training
Console.WriteLine("Beginning training using incremental back-propagation...");
nn.Train(normalizedTrainData.ToArray(), maxEpochs, learnRate, momentum, weightDecay);
Console.WriteLine("Done!");
Console.WriteLine();
#endregion
#region Results
double[] weights = nn.GetWeights();
Console.ForegroundColor = ConsoleColor.Green;
Console.WriteLine("Final neural network weights and bias values:");
Console.ResetColor();
Helpers.ShowVector(weights, 10, 3, true);
Console.WriteLine();
double trainAcc = nn.Accuracy(normalizedTrainData.ToArray());
Console.WriteLine("Accuracy on training data = " + trainAcc.ToString("F4"));
double testAcc = nn.Accuracy(normalizedTestData.ToArray());
Console.WriteLine("Accuracy on test data = " + testAcc.ToString("F4"));
Console.WriteLine();
//Console.ForegroundColor = ConsoleColor.Green;
//Console.WriteLine("Raw results:");
//Console.ResetColor();
//Console.WriteLine(nn.ToString());
#endregion
Console.ReadKey();
}