protected override void OnPaint(PaintEventArgs e)
{
var g = e.Graphics;
g.InterpolationMode = InterpolationMode.NearestNeighbor;
var entity = Connector.SelectedImage;//获得当前选择的贴图
var pos = CurrentLayer.Location;
if (!Config["MultipleLayer"].Boolean && entity?.Picture != null)
{
if (entity.Type == ColorBits.LINK && entity.Target != null)
{
entity = entity.Target;
}
var pictrue = entity.Picture;
var size = entity.Size.Star(Config["CanvasScale"].Decimal);
if (Config["LinearDodge"].Boolean)
{
pictrue = pictrue.LinearDodge();
}
if (Config["OnionSkin"].Boolean)
{
LastLayer?.Draw(g);
}
CurrentLayer.Tag = entity;
CurrentLayer.Size = size; //校正当前图层的宽高
CurrentLayer.Image = pictrue; //校正贴图
CurrentLayer.Draw(g); //绘制贴图
}
}
public double TrainToData(Vector input, Vector requiredOutput, double learningSpeed)
{
// Validate Input
if (input.Count != InputSize)
{
throw new ArgumentOutOfRangeException();
}
Vector result = Propergate(input);
LastLayer.Correction = requiredOutput - result;
// back propergate corrections
if (LayerCount == 0)
{
// easy
}
else
{
HiddernLayers.Last().Correction = LastLayer.BackPropergation;
for (int currectLayer = LayerCount - 2; currectLayer >= 0; currectLayer--)
{
HiddernLayers[currectLayer].Correction = HiddernLayers[currectLayer + 1].BackPropergation;
}
}
LastLayer.CorrectForError(learningSpeed);
HiddernLayers.ForEach(layer => layer.CorrectForError(learningSpeed));
return((requiredOutput - result).Average(correction => System.Math.Abs(correction)));
}
public void Forward(double[] input, double[] targets, bool doBackpropagation = true)
{
FirstLayer.Guess(input);
if (doBackpropagation)
{
//Console.WriteLine("Output in NeuralNetworkOld:");
//Console.WriteLine(LastLayer.O);
LastLayer.Backwards(targets);
}
}
protected Brain(SerializationInfo info, StreamingContext context)
{
//Get brains attributes
this.LayerCount = info.GetInt32("LayerCount");
// Input layer and output layer must be set later
//Get all the layers
Layer LastLayer = null;
for (int LayerLcv = this.LayerCount - 1; LayerLcv >= 0; LayerLcv--)
{
string LayerIdName = "Layer" + LayerLcv;
uint LayerUid = info.GetUInt32(LayerIdName + "Uid");
bool UseByteRes = info.GetBoolean(LayerIdName + "UseByteRes");
bool IsInput = info.GetBoolean(LayerIdName + "IsInput");
bool IsOutput = info.GetBoolean(LayerIdName + "IsOutput");
int NeuronCount = info.GetInt32(LayerIdName + "NeuronCount");
if (IsInput)
{
Layer NewLayer = new InputLayer(this, UseByteRes, LayerUid);
this.InLayer = (NewLayer as InputLayer);
NewLayer.DestinationLayer = LastLayer;
LastLayer.SourceLayer = NewLayer;
LastLayer = NewLayer;
}
else if (IsOutput)
{
LastLayer = new OutputLayer(this, UseByteRes, LayerUid);
this.OutLayer = (LastLayer as OutputLayer);
}
else
{
Layer NewLayer = new Layer(this, UseByteRes, LayerUid);
NewLayer.DestinationLayer = LastLayer;
LastLayer.SourceLayer = NewLayer;
LastLayer = NewLayer;
}
//Get all of this layers neurons
for (int NeuronLcv = 0; NeuronLcv < NeuronCount; NeuronLcv++)
{
string NeuronIdName = LayerIdName + "Neuron" + NeuronLcv;
double BiasWeight = info.GetDouble(NeuronIdName + "BiasWeight");
bool UseByte = info.GetBoolean(NeuronIdName + "UseByteResolution");
int OutConCount = info.GetInt32(NeuronIdName + "OutConCount");
uint NeuronUid = info.GetUInt32(NeuronIdName + "Uid");
Neuron NewNeuron = null;
if (IsInput)
{
NewNeuron = new InputNeuron(LastLayer, UseByte, BiasWeight, NeuronUid);
}
else if (IsOutput)
{
NewNeuron = new OutputNeuron(LastLayer, UseByte, BiasWeight, NeuronUid);
}
else
{
NewNeuron = new Neuron(LastLayer, UseByte, BiasWeight, NeuronUid);
}
LastLayer.AddNeuron(NewNeuron);
//Connect the new neuron to all its destinations
for (int SynapseLcv = 0; SynapseLcv < OutConCount; SynapseLcv++)
{
string SynapseIdName = NeuronIdName + "Synapse" + SynapseLcv;
uint ToUid = info.GetUInt32(SynapseIdName + "ToUid");
double Weight = info.GetDouble(SynapseIdName + "Weight");
Neuron ToConnectTo = this.FindNeuron(LastLayer, ToUid);
NewNeuron.ConnectToNeuron(ToConnectTo, Weight);
}
}
}
}
