支持向量回归（SVR）对函数的拟合近似

最新推荐文章于 2024-07-18 08:49:40 发布

qq_2773878606

最新推荐文章于 2024-07-18 08:49:40 发布

阅读量5.5k

点赞数

分类专栏：神经网络

本文链接：https://blog.csdn.net/qq_18343569/article/details/47111631

版权

神经网络专栏收录该内容

28 篇文章 12 订阅

订阅专栏

%
%  1D Regression example
%
close all
clear all
tic
%-------------------creating and plotting data----------------
n=50;
x=linspace(0,3,n)';
xx=linspace(0,3,n)';
xi=x;
yi=cos(exp(x));
y=cos(exp(xx));
figure(1)
hold on;
plot(xx,y,'g',xx,y,'b+');
%%
%----------------------Learning Parameters -------------------
C = 0.0001;  lambda = 0.000001; 
epsilon = .2;
kerneloption = 0.01;
kernel='gaussian';
verbose=1;
[xsup,ysup,w,w0] = svmreg(xi,yi,C,epsilon,kernel,kerneloption,lambda,verbose);
%---------------------Processing and plotting regression function--------

rx = svmval(xx,xsup,w,w0,kernel,kerneloption);
h = plot(xx,rx,'b'); 
set(h,'LineWidth',2);
h = plot(xsup,ysup,'or'); 
set(h,'LineWidth',2);
plot(xx,rx+epsilon,'b--'); 
plot(xx,rx-epsilon,'b--'); hold off;
title('Support Vector Machine Regression');
xlabel('x');ylabel('y');
hold off