24小时热门版块排行榜

返回列表

samggyy

金虫 (小有名气)

应助: 5 (幼儿园)
金币: 1457.4
散金: 10
红花: 1
帖子: 100
在线: 128.7小时
虫号: 483196
注册: 2007-12-26
性别: MM
专业: 模式识别

[求助] Matlab如何使用绘图画出3D SVM的decision boundary？

RT，Matlab中可以画出二维的SVM的decision boundary，代码如下，

load fisheriris;

features = meas(1:100,

;
featureSelcted = features(1:100,1:2);
groundTruthGroup = species(1:100);

svmStruct = svmtrain(featureSelcted, groundTruthGroup, ...
'Kernel_Function', 'rbf', 'boxconstraint', Inf, 'showplot', true, 'Method', 'QP');
svmClassified = svmclassify(svmStruct,featureSelcted,'showplot',true);

如果使用三维的特征，即把代码改为featureSelcted = features(1:100,1:3); 无法出一个三维的超平面了。思路可能是：

cubeXMin = min(featureSelcted(:,1))-0.5;
cubeYMin = min(featureSelcted(:,2))-0.5;
cubeZMin = min(featureSelcted(:,3))-0.5;

cubeXMax = max(featureSelcted(:,1))+0.5;
cubeYMax = max(featureSelcted(:,2))+0.5;
cubeZMax = max(featureSelcted(:,3))+0.5;

cubeMesh = meshgrid(cubeXMin:0.5:cubeXMax,cubeYMin:0.5:cubeYMax,cubeZMin:0.5:cubeZMax);

sv = svmStruct.SupportVectors;
alphaHat = svmStruct.Alpha;
bias = svmStruct.Bias;
kfun = svmStruct.KernelFunction;
kfunargs = svmStruct.KernelFunctionArgs;

f = (feval(kfun,sv,cubeMesh,kfunargs{:})'*alphaHat(

) + bias;

decisionBoundary = sign(f);
decisionBoundaryInd = find(decisionBoundary==0);

figure,
scatter3(featureSelcted(:,1),featureSelcted(:,2),featureSelcted(:,3));
surf(decisionBoundaryInd);

但是不知道怎么完成，请教有没有人知道怎么做的。谢谢！