public void Neural_Network_Study_Test()
{
var data = new[] {
new { Study = 2.0, Beer = 3.0, Passed = false},
new { Study = 3.0, Beer = 4.0, Passed = false},
new { Study = 1.0, Beer = 6.0, Passed = false},
new { Study = 4.0, Beer = 5.0, Passed = false},
new { Study = 6.0, Beer = 2.0, Passed = true},
new { Study = 8.0, Beer = 3.0, Passed = true},
new { Study = 12.0, Beer = 1.0, Passed = true},
new { Study = 3.0, Beer = 2.0, Passed = true}
};
var descriptor = Descriptor.New("Student")
.With("Study").As(typeof(double))
.With("Beer").As(typeof(double))
.Learn("Passed").As(typeof(bool));
NeuralNetworkGenerator generator = new NeuralNetworkGenerator()
{
Descriptor = descriptor,
NormalizeFeatures = true,
LearningRate = 0.1
};
var model = Learner.Learn(data, 0.8, 10, generator).Model;
var test = model.PredictValue(new { Study = 7.0, Beer = 1.0, Passed = false });
Assert.Equal(true, test);
}
public void xor_test_learner()
{
var xor = new[]
{
new { a = false, b = false, c = false },
new { a = false, b = true, c = true },
new { a = true, b = false, c = true },
new { a = true, b = true, c = false },
new { a = false, b = false, c = false },
new { a = false, b = true, c = true },
new { a = true, b = false, c = true },
new { a = true, b = true, c = false },
new { a = false, b = false, c = false },
new { a = false, b = true, c = true },
new { a = true, b = false, c = true },
new { a = true, b = true, c = false },
new { a = false, b = false, c = false },
new { a = false, b = true, c = true },
new { a = true, b = false, c = true },
new { a = true, b = true, c = false },
new { a = false, b = false, c = false },
new { a = false, b = true, c = true },
new { a = true, b = false, c = true },
new { a = true, b = true, c = false },
};
var d = Descriptor.New("XOR")
.With("a").As(typeof(bool))
.With("b").As(typeof(bool))
.Learn("c").As(typeof(bool));
var generator = new NeuralNetworkGenerator { Descriptor = d };
var model = Learner.Learn(xor, .75, 1000, generator).Model;
Matrix x = new[,]
{{ -1, -1 }, // false, false -> -
{ -1, 1 }, // false, true -> +
{ 1, -1 }, // true, false -> +
{ 1, 1 }}; // true, true -> -
Vector y = new[] { 0, 0, 0, 0 };
for (int i = 0; i < x.Rows; i++)
y[i] = model.Predict(x[i, VectorType.Row]);
}
public void XOR_Test_Learner()
{
var xor = new[]
{
new { a = false, b = false, c = false },
new { a = false, b = true, c = true },
new { a = true, b = false, c = true },
new { a = true, b = true, c = false },
new { a = false, b = false, c = false },
new { a = false, b = true, c = true },
new { a = true, b = false, c = true },
new { a = true, b = true, c = false },
new { a = false, b = false, c = false },
new { a = false, b = true, c = true },
new { a = true, b = false, c = true },
new { a = true, b = true, c = false },
new { a = false, b = false, c = false },
new { a = false, b = true, c = true },
new { a = true, b = false, c = true },
new { a = true, b = true, c = false },
new { a = false, b = false, c = false },
new { a = false, b = true, c = true },
new { a = true, b = false, c = true },
new { a = true, b = true, c = false },
};
var d = Descriptor.New("XOR")
.With("a").As(typeof(bool))
.With("b").As(typeof(bool))
.Learn("c").As(typeof(bool));
var generator = new NeuralNetworkGenerator { Descriptor = d };
var model = Learner.Learn(xor, .75, 10, generator).Model;
Matrix x = new[,]
{{ -1, -1 }, // false, false -> -
{ -1, 1 }, // false, true -> +
{ 1, -1 }, // true, false -> +
{ 1, 1 }}; // true, true -> -
Vector actual = new int[] { -1, 1, 1, -1 };
Vector y = new[] { 0, 0, 0, 0 };
for (int i = 0; i < x.Rows; i++)
y[i] = model.Predict(x[i, VectorType.Row]);
var score = numl.Supervised.Score.ScorePredictions(y, actual);
Console.WriteLine(string.Format("Neural Network XOR Test (score) =>\n{0}", score.RMSE));
}
