public void return_0_as_error_if_it_is_between_the_acceptance_matcher_limits()
{
var neuralNetwork = new NeuralNetworkBuilder(new ConnectionProperties(_weightGenerator),
new PerceptronProperties(_thresholdGenerator),
new AcceptanceMatcher(-0.049, 0.049))
.WithLayer(1).From(1).To(1)
.Build();
neuralNetwork.ExitValues = new ValueList<double> { 2.049 };
neuralNetwork.ExpectedExitValues = new ValueList<double> { 2 };
neuralNetwork.GetErrorForExit(1).Should().Be(0.0);
}
public void return_0_point_5_when_it_has_no_hidden_layers_entry_value_is_0_threshold_is_0_and_weight_is_1()
{
_thresholdGenerator.Generate().Returns(0.0);
_weightGenerator.Generate().Returns(1.0);
var neuralNetwork = new NeuralNetworkBuilder(new ConnectionProperties(_weightGenerator),
new PerceptronProperties(_thresholdGenerator),
new AcceptanceMatcher(0.0, 0.0))
.WithLayer(1).From(1).To(1)
.Build();
neuralNetwork.EntryValues = new ValueList<double> { 0.0 };
neuralNetwork.Execute();
neuralNetwork.ExitValues[1].Should().Be(0.5);
}
public void return_0_point_26894142_when_it_has_no_hidden_layers_entry_values_are_2_and_4_threshold_is_minus7_and_weights_are_1()
{
_thresholdGenerator.Generate().Returns(-7.0);
_weightGenerator.Generate().Returns(1.0);
var neuralNetwork = new NeuralNetworkBuilder(new ConnectionProperties(_weightGenerator),
new PerceptronProperties(_thresholdGenerator),
new AcceptanceMatcher(0.0, 0.0))
.WithLayer(1).From(2).To(1)
.Build();
neuralNetwork.EntryValues = new ValueList<double> { 2.0, 4.0 };
neuralNetwork.Execute();
neuralNetwork.ExitValues[1].Should().BeApproximately(0.26894142, 0.00000001);
}
public void return_an_exit_value_of_0_when_perceptrons_have_a_very_low_threshold()
{
_thresholdGenerator.Generate().Returns(-9999.0);
_weightGenerator.Generate().Returns(1.0);
var neuralNetwork = new NeuralNetworkBuilder(new ConnectionProperties(_weightGenerator),
new PerceptronProperties(_thresholdGenerator),
new AcceptanceMatcher(0.0, 0.0))
.WithLayer(1).From(1).To(1)
.WithLayer(2).From(1).To(1)
.Build();
neuralNetwork.Execute();
neuralNetwork.ExitValues[1].Should().Be(0.0);
}
public void return_correct_modified_weight_for_a_single_layer()
{
_thresholdGenerator.Generate().Returns(0.0);
_weightGenerator.Generate().Returns(1.0);
var neuralNetwork = new NeuralNetworkBuilder(new ConnectionProperties(_weightGenerator),
new PerceptronProperties(_thresholdGenerator),
new AcceptanceMatcher(0.0, 0.0))
.WithLayer(1).From(2).To(2)
.WithLayer(2).From(2).To(2)
.WithLayer(3).From(2).To(2)
.WithLayer(4).From(2).To(2)
.Build();
neuralNetwork.EntryValues = new ValueList<double> { 0.2, 0.4, 0.6, 0.8 };
neuralNetwork.ExpectedExitValues = new ValueList<double> { 0.9, 0.6, 0.3, 0.0 };
neuralNetwork.Execute();
neuralNetwork.ExecuteBackPropagation();
}
public void return_correct_error_value_for_an_exit_with_an_associated_expected_value()
{
var neuralNetwork = new NeuralNetworkBuilder(new ConnectionProperties(_weightGenerator),
new PerceptronProperties(_thresholdGenerator),
new AcceptanceMatcher(0.0, 0.0))
.WithLayer(1).From(1).To(1)
.Build();
neuralNetwork.EntryValues = new ValueList<double> { 1.0 };
neuralNetwork.ExitValues = new ValueList<double> { 0.5 };
neuralNetwork.ExpectedExitValues = new ValueList<double> { 0.8 };
neuralNetwork.GetErrorForExit(1).Should().BeApproximately(-0.30000000, 0.00000001);
}