private static Vector <double> PredictOneVsAll(Matrix <double> all_theta, Matrix <double> X)
{
int m = X.RowCount;
// Add ones to the X data matrix
X = X.InsertColumn(0, Vector <double> .Build.Dense(m, 1.0));
Vector <double> pred = Vector <double> .Build.Dense(m);
var h = LogisticRegression.Sigmoid(X * all_theta.Transpose());
for (int i = 0; i < m; i++)
{
pred[i] = h.Row(i).MaximumIndex() + 1;
}
return(pred);
}
static void Main(string[] args)
{
if (!System.Console.IsOutputRedirected)
{
System.Console.Clear();
}
CultureInfo.CurrentCulture = CultureInfo.CreateSpecificCulture("en-US");
System.Console.WriteLine("Multi-class Classification and Neural Networks ex.3");
System.Console.WriteLine("================================================\n");
var M = Matrix <double> .Build;
var V = Vector <double> .Build;
// read all matrices of a file by name into a dictionary
Dictionary <string, Matrix <double> > ms = MatlabReader.ReadAll <double>("data\\ex3data1.mat");
Matrix <double> X = ms["X"];
Vector <double> y = ms["y"].Column(0);
// get a casual sequence of 100 int numbers
var srs = new MathNet.Numerics.Random.SystemRandomSource();
var seq = srs.NextInt32Sequence(0, 5000).Take(100).ToList();
// Randomly select 100 data points to display
Vector <double>[] sel = new Vector <double> [100];
int idx = 0;
Vector <double> v = V.Dense(400);
foreach (int i in seq)
{
sel[idx++] = X.Row(i);
}
// display
DisplayData(sel);
Pause();
//// ============ Part 2a: Vectorize Logistic Regression ============
// In this part of the exercise, you will reuse your logistic regression
// code from the last exercise. You task here is to make sure that your
// regularized logistic regression implementation is vectorized. After
// that, you will implement one-vs-all classification for the handwritten
// digit dataset.
//
// Test case for lrCostFunction
System.Console.WriteLine("\nTesting Cost Function with regularization");
Vector <double> theta_t = V.DenseOfArray(new[] { -2.0, -1, 1, 2 });
Matrix <double> X_t = M.DenseOfArray(new [, ] {
{ 1.0, 0.1, 0.6, 1.1 },
{ 1.0, 0.2, 0.7, 1.2 },
{ 1.0, 0.3, 0.8, 1.3 },
{ 1.0, 0.4, 0.9, 1.4 },
{ 1.0, 0.5, 1.0, 1.5 },
});
Vector <Double> y_t = V.DenseOfArray(new [] { 1.0, 0, 1, 0, 1 });
int lambda_t = 3;
LogisticRegression lr = new LogisticRegression(X_t, y_t);
lr.Lambda = lambda_t;
double J = lr.Cost(theta_t);
Vector <double> grad = lr.Gradient(theta_t);
System.Console.WriteLine("\nCost: {0:f5}\n", J);
System.Console.WriteLine("Expected cost: 2.534819\n");
System.Console.WriteLine("Gradients:\n");
System.Console.WriteLine(" {0:f5} \n", grad);
System.Console.WriteLine("Expected gradients:\n");
System.Console.WriteLine(" 0.146561\n -0.548558\n 0.724722\n 1.398003\n");
Pause();
//// ============ Part 2b: One-vs-All Training ============
System.Console.WriteLine("\nTraining One-vs-All Logistic Regression...\n");
double lambda = 0.1;
int num_labels = 10;
Matrix <double> all_theta = OneVsAll(X, y, num_labels, lambda);
Pause();
// ================ Part 3: Predict for One-Vs-All ================
Vector <double> pred = PredictOneVsAll(all_theta, X);
Vector <double> comp = V.Dense(y.Count);
for (int i = 0; i < y.Count; i++)
{
if (pred[i] == y[i])
{
comp[i] = 1;
}
else
{
comp[i] = 0;
}
}
double accuracy = comp.Mean() * 100;
System.Console.WriteLine("\nTraining Set Accuracy: {0:f5}\n", accuracy);
Pause();
}
