public AxonGene GetGenes()
{
return(new AxonGene
{
ActivationFunction = _activationFunction.GetType(),
Weights = _terminals.Select(d => d.Weight).ToList()
});
}
private String MakeActivationFunctionString(IActivationFunction act)
{
StringBuilder result = new StringBuilder();
result.Append(act.GetType().Name);
for (int i = 0; i < act.Params.Length; i++)
{
result.Append('|');
result.Append(CSVFormat.EgFormat.Format(act.Params[i],
EncogFramework.DefaultPrecision));
}
return(result.ToString());
}
private void updateGraph()
{
Type functionType = cbActivationFunction.SelectedItem as Type;
if (functionType != null && (function == null || functionType != function.GetType()))
{
this.function = (IActivationFunction)Activator.CreateInstance(functionType);
this.propertyGrid.SelectedObject = function;
this.activationFunctionView.Function = function;
this.activationFunctionView.Domain = new DoubleRange(-1.0, 1.0);
this.activationFunctionView.Steps = 50;
this.activationFunctionView.Plot();
}
}
public void Plot()
{
if (this.function == null)
{
throw new InvalidOperationException();
}
zedGraphControl.GraphPane.CurveList.Clear();
double x, y;
double stepSize = plotDomain.Length / plotSteps;
PointPairList list = new PointPairList();
for (int i = 0; i < plotSteps; i++)
{
x = this.plotDomain.Min + ((double)i * stepSize);
switch (this.derivative)
{
case FunctionDerivative.None:
y = this.function.Function(x); break;
case FunctionDerivative.First:
y = this.function.Derivative(x); break;
case FunctionDerivative.Second:
y = this.function.Derivative2(x); break;
default:
goto case FunctionDerivative.None;
}
list.Add(x, y);
}
LineItem curve = zedGraphControl.GraphPane.
AddCurve(function.GetType().Name, list, Color.Red, SymbolType.None);
//curve.Line.IsSmooth = true;
//curve.Line.SmoothTension = 0.1f;
curve.Line.IsAntiAlias = true;
curve.Line.Width = 2f;
zedGraphControl.AxisChange();
zedGraphControl.Invalidate();
}
/// <summary>
/// Write a property as an activation function.
/// </summary>
///
/// <param name="name">The name of the property.</param>
/// <param name="act">The activation function.</param>
public void WriteProperty(String name,
IActivationFunction act)
{
var result = new StringBuilder();
result.Append(act.GetType().Name);
for (int i = 0; i < act.Params.Length; i++)
{
result.Append('|');
result.Append(CSVFormat.EgFormat.Format(act.Params[i],
EncogFramework.DefaultPrecision));
}
WriteProperty(name, result.ToString());
}
/// <summary>
/// Save the activation function.
/// </summary>
/// <param name="activationFunction">The activation function.</param>
/// <param name="xmlOut">The XML.</param>
public static void SaveActivationFunction(
IActivationFunction activationFunction, WriteXML xmlOut)
{
if (activationFunction != null)
{
xmlOut.BeginTag(BasicLayerPersistor.TAG_ACTIVATION);
xmlOut.BeginTag(activationFunction.GetType().Name);
String[] names = activationFunction.ParamNames;
for (int i = 0; i < names.Length; i++)
{
String str = names[i];
double d = activationFunction.Params[i];
xmlOut.AddAttribute(str, "" + CSVFormat.EG_FORMAT.Format(d, 10));
}
xmlOut.EndTag();
xmlOut.EndTag();
}
}
private String ToSingleLineArray(
IActivationFunction[] activationFunctions)
{
var result = new StringBuilder();
result.Append('[');
for (int i = 0; i < activationFunctions.Length; i++)
{
if (i > 0)
{
result.Append(',');
}
IActivationFunction af = activationFunctions[i];
if (af is ActivationSigmoid)
{
result.Append("ENCOG.ActivationSigmoid.create()");
}
else if (af is ActivationTANH)
{
result.Append("ENCOG.ActivationTANH.create()");
}
else if (af is ActivationLinear)
{
result.Append("ENCOG.ActivationLinear.create()");
}
else if (af is ActivationElliott)
{
result.Append("ENCOG.ActivationElliott.create()");
}
else if (af is ActivationElliott)
{
result.Append("ENCOG.ActivationElliott.create()");
}
else
{
throw new AnalystCodeGenerationError(
"Unsupported activatoin function for code generation: "
+ af.GetType().Name);
}
}
result.Append(']');
return(result.ToString());
}
//Methods
private void TestActivation(IActivationFunction af, int simLength, double constCurrent, int from, int count)
{
for (int i = 1; i <= simLength; i++)
{
double signal;
double input;
if (i >= from && i < from + count)
{
input = double.IsNaN(constCurrent) ? _rand.NextDouble() : constCurrent;
}
else
{
input = 0d;
}
signal = af.Compute(input);
Console.WriteLine($"{af.GetType().Name} step {i}, State {(af.TypeOfActivation == ActivationType.Spiking ? af.InternalState : signal)} signal {signal}");
}
Console.ReadLine();
return;
}
public override List <string> DescribeSelf()
{
List <String> output = new List <string>();
var pair = states.FirstOrDefault();
output.Add("Classification using neural network [" + pair.Value.machine.GetType().Name + "] with " + pair.Value.machine.Layers.Count() + " layers.");
output.Add("Supervised learning [" + teacherRef.GetType().Name + "] with LearningRate = " + setup.neuralnetwork.learningRate + " and Momentum = " + setup.neuralnetwork.momentum + ".");
output.Add("Learning in max. iterations [" + setup.neuralnetwork.learningIterationsMax + "], terminated earlier if error rate is lower then [" + setup.neuralnetwork.errorLowerLimit.ToString("F5") + "]");
// output.Add("Input layer [0] -> [" + pair.Value.data.NumberOfInputs + "]");
for (int i = 0; i < pair.Value.machine.Layers.Length; i++)
{
Layer l = pair.Value.machine.Layers[i];
output.Add("Layer [" + (i) + "] -> In[" + l.InputsCount + "] -> Neurons[" + l.Neurons.Length + "] -> Out[" + l.Output.Length + "]");
}
output.Add("Output layer [" + (pair.Value.machine.Layers.Length) + "] -> [" + pair.Value.data.NumberOfClasses + "]");
output.Add("Neuron function [" + activationFunction.GetType().Name + "]. Trained with [" + pair.Value.data.NumberOfCases + "] cases.");
return(output);
}
/// <summary>
/// Save the activation function.
/// </summary>
/// <param name="activationFunction">The activation function.</param>
/// <param name="xmlOut">The XML.</param>
public static void SaveActivationFunction(
IActivationFunction activationFunction, WriteXML xmlOut)
{
if (activationFunction != null)
{
xmlOut.BeginTag(BasicLayerPersistor.TAG_ACTIVATION);
xmlOut.BeginTag(activationFunction.GetType().Name);
String[] names = activationFunction.ParamNames;
for (int i = 0; i < names.Length; i++)
{
String str = names[i];
double d = activationFunction.Params[i];
xmlOut.AddAttribute(str, "" + CSVFormat.EG_FORMAT.Format(d, 10));
}
xmlOut.EndTag();
xmlOut.EndTag();
}
}
private String MakeActivationFunctionString(IActivationFunction act)
{
StringBuilder result = new StringBuilder();
result.Append(act.GetType().Name);
for (int i = 0; i < act.Params.Length; i++)
{
result.Append('|');
result.Append(CSVFormat.EgFormat.Format(act.Params[i],
EncogFramework.DefaultPrecision));
}
return result.ToString();
}
/// <summary>
/// Write a property as an activation function.
/// </summary>
///
/// <param name="name">The name of the property.</param>
/// <param name="act">The activation function.</param>
public void WriteProperty(String name,
IActivationFunction act)
{
var result = new StringBuilder();
result.Append(act.GetType().Name);
for (int i = 0; i < act.Params.Length; i++)
{
result.Append('|');
result.Append(CSVFormat.EgFormat.Format(act.Params[i],
EncogFramework.DefaultPrecision));
}
WriteProperty(name, result.ToString());
}
private void CanItLearnRulesWith(IList<IMLDataPair> inputData, IList<IMLDataPair> verfData, int hiddenLayerCount, int neuronCount, IActivationFunction actFunc, double learnRate, double momentum, int batchSize, int maxEpochs)
{
var model = new DbModel();
var funcName = actFunc.GetType().Name;
var tdCount = inputData.Count();
if (model.TicTacToeResult.Any(r => r.HiddenLayerCount == hiddenLayerCount &&
r.NeuronPerLayercount == neuronCount &&
r.ActivationFunction == funcName &&
r.LearningRate == learnRate &&
r.BatchSize == batchSize &&
r.Momentum == momentum &&
r.Name == Name &&
r.Epochs == maxEpochs &&
r.TrainingDataCount == tdCount))
return;
var nn = CreateNetwork(inputData, hiddenLayerCount, neuronCount, actFunc);
var train = new Backpropagation(nn, new BasicMLDataSet(inputData), learnRate, momentum);
train.BatchSize = batchSize;
int epoch = 1;
do
{
train.Iteration();
epoch++;
} while (epoch < maxEpochs);
int good = verfData.Count(verf => { var output = nn.Compute(verf.Input); return Enumerable.Range(0, 9).All(i => Math.Round(output[i]) == Math.Round(verf.Ideal[i])); });
int bad = VerfDataCount - good;
var result = new TicTacToeResult()
{
HiddenLayerCount = hiddenLayerCount,
NeuronPerLayercount = neuronCount,
ActivationFunction = funcName,
Bad = bad,
Good = good,
TrainingDataCount = tdCount,
Momentum = momentum,
LearningRate = learnRate,
BatchSize = batchSize,
Epochs = epoch,
Error = train.Error,
Name = Name,
};
model.TicTacToeResult.Add(result);
model.SaveChanges();
}
public void WriteProperty(string name, IActivationFunction act)
{
StringBuilder builder = new StringBuilder();
builder.Append(act.GetType().Name);
int index = 0;
goto Label_0030;
Label_0011:
builder.Append(CSVFormat.EgFormat.Format(act.Params[index], 10));
index++;
Label_0030:
if (index < act.Params.Length)
{
builder.Append('|');
goto Label_0011;
}
if ((((uint) index) | 3) == 0)
{
goto Label_0011;
}
this.WriteProperty(name, builder.ToString());
}
public LayerUnit(int length, IActivationFunction actfunc)
{
var q = from i in Enumerable.Range(0, length)
select new NeuronUnit
{
sign = 1.0f,
activation = 0.0f,
bias = UnityEngine.Random.value,
af = actfunc != null ? (IActivationFunction)Activator.CreateInstance(actfunc.GetType()) : null
}
;
this.neurons = q.ToArray();
}
//Methods
/// <summary>
/// Builds name of the specified activation function
/// </summary>
/// <param name="activationCfg">Activation function configuration</param>
private string GetActivationName(RCNetBaseSettings activationCfg)
{
IActivationFunction aFn = ActivationFactory.Create(activationCfg, _rand);
return(aFn.TypeOfActivation.ToString() + "-" + aFn.GetType().Name.Replace("Settings", string.Empty));
}