neuralnode.cpp

使用神经网络开发包实现图形化的神经网络模拟

// NeuronNode.cpp: implementation of the NeuronNode class.
//
//////////////////////////////////////////////////////////////////////

#include "stdafx.h"
#include "NeuralNetwork.h"
#include "NeuralNode.h"

#ifdef _DEBUG
#undef THIS_FILE
static char THIS_FILE[]=__FILE__;
#define new DEBUG_NEW
#endif

//////////////////////////////////////////////////////////////////////
// Construction/Destruction
//////////////////////////////////////////////////////////////////////
IMPLEMENT_SERIAL(NeuralNode, CObject, 1)
NeuralNode::NeuralNode()
{
	for(int i=0;i<EDGENUM;i++)
	{
		edgeType[i]=0;
	}
	//======初始化偏置值========
	bias=0.0;
	//======初始化传输函数======
	activeFunction=gaussian;
}

NeuralNode::~NeuralNode()
{

}

NeuralNode::NeuralNode(CPoint pos,int index)
{
	//======初始化传输函数======
	activeFunction=gaussian;
	//======初始化偏置值========
	bias=0.0;
	//======初始化节点坐标======
	nodePos=pos;
	//======初始化节点索引======
	totalIndex=index;
	//=====初始化边缘节点类型======
	for(int i=0;i<EDGENUM;i++)
	{
		edgeType[i]=0;
	}
	//======初始化中心点矩形区域======
	centerPos=CRect(pos.x+3*LINESIZE,pos.y+3*LINESIZE,pos.x+6*LINESIZE,pos.y+6*LINESIZE);
	//======初始化边缘节点矩形区域=======
	edgeNodePos[0]=CRect(pos.x+3*LINESIZE,pos.y,pos.x+4*LINESIZE,pos.y+LINESIZE);
	edgeNodePos[1]=CRect(pos.x+5*LINESIZE,pos.y,pos.x+6*LINESIZE,pos.y+LINESIZE);

	edgeNodePos[2]=CRect(pos.x+LINESIZE,pos.y+LINESIZE,pos.x+2*LINESIZE,pos.y+2*LINESIZE);
	edgeNodePos[3]=CRect(pos.x+7*LINESIZE,pos.y+LINESIZE,pos.x+8*LINESIZE,pos.y+2*LINESIZE);

	edgeNodePos[4]=CRect(pos.x,pos.y+3*LINESIZE,pos.x+LINESIZE,pos.y+4*LINESIZE);
	edgeNodePos[5]=CRect(pos.x+8*LINESIZE,pos.y+3*LINESIZE,pos.x+9*LINESIZE,pos.y+4*LINESIZE);

	edgeNodePos[6]=CRect(pos.x,pos.y+5*LINESIZE,pos.x+LINESIZE,pos.y+6*LINESIZE);
	edgeNodePos[7]=CRect(pos.x+8*LINESIZE,pos.y+5*LINESIZE,pos.x+9*LINESIZE,pos.y+6*LINESIZE);

	edgeNodePos[8]=CRect(pos.x+LINESIZE,pos.y+7*LINESIZE,pos.x+2*LINESIZE,pos.y+8*LINESIZE);
	edgeNodePos[9]=CRect(pos.x+7*LINESIZE,pos.y+7*LINESIZE,pos.x+8*LINESIZE,pos.y+8*LINESIZE);

	edgeNodePos[10]=CRect(pos.x+3*LINESIZE,pos.y+8*LINESIZE,pos.x+4*LINESIZE,pos.y+9*LINESIZE);
	edgeNodePos[11]=CRect(pos.x+5*LINESIZE,pos.y+8*LINESIZE,pos.x+6*LINESIZE,pos.y+9*LINESIZE);
}

void NeuralNode::setBias(real newBias)
{
	bias=newBias;
}

void NeuralNode::setActivationFunction(ActivationFunction newFunction)
{
	activeFunction=newFunction;
}

void NeuralNode::setWeight(real newWeight, int btnPosType)
{
	outputLine[btnPosType].setWeight(newWeight);
}

ActivationFunction NeuralNode::getActivationFunction()
{
	return activeFunction;
}

void NeuralNode::Serialize(CArchive &ar)
{
	if(ar.IsStoring())
	{
		ar<<bias<<centerPos<<layerIndex<<neuronIndex<<nodePos<<totalIndex;//===<<activeFunction
		for(int i=0;i<EDGENUM;i++)
		{
			ar<<edgeNodePos[i]<<edgeType[i]<<inputNeuronIndex[i];
		}
	}
	else
	{
		ar>>bias>>centerPos>>layerIndex>>neuronIndex>>nodePos>>totalIndex;//=====>>activeFunction
		for(int i=0;i<EDGENUM;i++)
		{
			ar>>edgeNodePos[i]>>edgeType[i]>>inputNeuronIndex[i];
		}
	}
	for(int i=0;i<EDGENUM;i++)
	{
		outputLine[i].Serialize(ar);
	}
}
NeuralNode& NeuralNode::operator =(const NeuralNode &s)
{
	activeFunction=s.activeFunction;
	bias=s.bias;
	centerPos=s.centerPos;
	for(int i=0;i<EDGENUM;i++)
	{
		edgeNodePos[i]=s.edgeNodePos[i];
		inputNeuronIndex[i]=s.inputNeuronIndex[i];
		outputLine[i]=s.outputLine[i];
		edgeType[i]=s.edgeType[i];
	}
	layerIndex=s.layerIndex;
	neuronIndex=s.neuronIndex;
	nodePos=s.nodePos;
	totalIndex=s.totalIndex;
	return *this;
}

NeuralNode::NeuralNode( const NeuralNode &s )
{
	activeFunction=s.activeFunction;
	bias=s.bias;
	centerPos=s.centerPos;
	for(int i=0;i<EDGENUM;i++)
	{
		edgeNodePos[i]=s.edgeNodePos[i];
		inputNeuronIndex[i]=s.inputNeuronIndex[i];
		outputLine[i]=s.outputLine[i];
		edgeType[i]=s.edgeType[i];
	}
	layerIndex=s.layerIndex;
	neuronIndex=s.neuronIndex;
	nodePos=s.nodePos;
	totalIndex=s.totalIndex;
}