chapter 0 —— 半月形正负样本

源程序

《机器学习与神经网络》一书的代码中有生成半月形数据用于训练与测试。我将原matlab代码转为C/C++的代码，方便后续程序。

原matlab代码如下：

function [data, data_shuffled] = halfmoon(rad,width,d,n_samp)
% A function to generate the halfmoon data
% where Input:
%         rad  - central radius of the half moon
%        width - width of the half moon
%           d  - distance between two half moon
%      n_samp  - total number of the samples
%       Output:
%         data - output data
%data_shuffled - shuffled data
% For example
% halfmoon(10,2,0,1000) will generate 1000 data of 
% two half moons with radius [9-11] and space 0.

% Copyright
% Yanbo Xue
% Adaptive Systems Laboratory
% McMaster University
% yxue@soma.mcmaster.ca
% Feb. 22, 2007

if rad < width/2,
    error('The radius should be at least larger than half the width');
end

if mod(n_samp,2)~=0,
    error('Please make sure the number of samples is even');
end

aa = rand(2,n_samp/2);
radius = (rad-width/2) + width*aa(1,:);
theta = pi*aa(2,:);


x     = radius.*cos(theta);
y     = radius.*sin(theta);
label = 1*ones(1,length(x));  % label for Class 1

x1    = radius.*cos(-theta) + rad;
y1    = radius.*sin(-theta) - d;
label1= -1*ones(1,length(x)); % label for Class 2

data  = [x, x1;
         y, y1;
         label, label1];
     
[n_row, n_col] = size(data);

shuffle_seq = randperm(n_col);

for i = (1:n_col),
    data_shuffled(:,i) = data(:,shuffle_seq(i));
end;

C/C++程序

// moon_generator.cpp : Defines the entry point for the console application.
//

int half_moon_samples(int rad,int width,int distance,int num_samp)
{
	int i,j;
	double **aa=new double*[2];
	double **data=new double*[3];
	double *radius=new double[num_samp/2];
	double *theta=new double[num_samp/2];

	double *x1=new double[num_samp/2];
	double *y1=new double[num_samp/2];
	double *label1=new double[num_samp/2];

	double *x2=new double[num_samp/2];
	double *y2=new double[num_samp/2];
	double *label2=new double[num_samp/2];

	srand(time(0));    //置随机数种子。
	if (rad<width/2)
	{
		cout<<"Error! The radius should be at least larger than half the width."<<endl;	
		return -1;
	}

	if (num_samp % 2!=0)
	{
		cout<<"Please make sure the number of samples is even."<<endl;	
		return -1;
	}

	for(i=0;i<2;i++)
	{
		aa[i]=new double[num_samp/2];
	}

	for (i=0;i<3;i++)
	{
		data[i]=new double[num_samp];
	}

	for (i=0;i<2;i++)
	{
		for (j=0;j<num_samp/2;j++)
		{

			aa[i][j]=rand()%100*0.01;        //取得整数1到x之间的随机数。
			if (i==0)
			{
				radius[j]=(rad-width/2)+width*aa[i][j];
			}
			else
			{
				theta[j]=3.14*aa[i][j];

				x1[j]=radius[j]*cos(theta[j]);
				y1[j]=radius[j]*sin(theta[j]);
				label1[j]=1;

				x2[j]=radius[j]*cos(-theta[j])+rad;
				y2[j]=radius[j]*sin(-theta[j])-distance;
				label2[j]=-1;
			}
		}
	}

	for (i=0;i<num_samp;i++)
	{
		if (i<num_samp/2)
		{
			data[0][i]=x1[i];
			data[1][i]=y1[i];
			data[2][i]=label1[i];
		}
		else
		{
			data[0][i]=x2[i-num_samp/2];
			data[1][i]=y2[i-num_samp/2];
			data[2][i]=label2[i-num_samp/2];
		}
	}

	/*****************随机洗牌(K次两两交换),对data中的数据进行重新分配*******************/

	double temp[3];
	int K=6000;
	int p1,p2;
	for (i=0;i<K;i++)
	{
		p1=rand()%(num_samp-1);
		p2=rand()%(num_samp-1);

		temp[0]=data[0][p1];
		temp[1]=data[1][p1];
		temp[2]=data[2][p1];

		data[0][p1]=data[0][p2];
		data[1][p1]=data[1][p2];
		data[2][p1]=data[2][p2];

		data[0][p2]=temp[0];
		data[1][p2]=temp[1];
		data[2][p2]=temp[2];		
	}
	/*****************将数据输出到文本*******************/

	ofstream fout;
	fout.open("data.txt");

	for (i=0;i<num_samp;i++)
	{
		fout<<data[2][i]<<" "<<data[0][i]<<" "<<data[1][i]<<endl;
	}
	fout.close();

	return 0;
}

结果

将生成的data.txt用matlab程序plot出的结果为：

其中，半径=10，宽度=6，距离=1，总样本数=3000。

代码下载

http://download.csdn.net/detail/tyj88tcai/6834253