Thread Subject: svm classification ---- distance to the classifying hyperplane (surface, curve)

Hi,

I having been using matlab svmtrain/svmclassify to classify my two-dimensional data with rbf function in the training process. Currently, svmclassify only outputs the class labels. However, I also would like to know how far the classified data points are from the classifying hyperplane (or curve in the original 2-dim plane in my case) to sort of measure the confidence in classifying each data point. I would appreciate it if anyone know how to make svmclassify out such infos.

"Guodong " <liug@pitt.edu> wrote in message <h1thnq$p62$1@fred.mathworks.com>...
> Hi,
>
> I having been using matlab svmtrain/svmclassify to classify my two-dimensional data with rbf function in the training process. Currently, svmclassify only outputs the class labels. However, I also would like to know how far the classified data points are from the classifying hyperplane (or curve in the original 2-dim plane in my case) to sort of measure the confidence in classifying each data point. I would appreciate it if anyone know how to make svmclassify out such infos.


that's what i am thinking about as well

"Arthur Zheng" <hzheng7@gatech.edu> wrote in message <h9c0oa$er7$1@fred.mathworks.com>...
> "Guodong " <liug@pitt.edu> wrote in message <h1thnq$p62$1@fred.mathworks.com>...
> > Hi,
> >
> > I having been using matlab svmtrain/svmclassify to classify my two-dimensional data with rbf function in the training process. Currently, svmclassify only outputs the class labels. However, I also would like to know how far the classified data points are from the classifying hyperplane (or curve in the original 2-dim plane in my case) to sort of measure the confidence in classifying each data point. I would appreciate it if anyone know how to make svmclassify out such infos.
>
>
> that's what i am thinking about as well

The distance of x from the hyperplan (in the *feature* space) - scaled by the margin - is:

d(x) = | sum K(xi,x)*(alphai.*yi) - 0.5*sum K(xi,s)*(alphai.*yi) |

xi is training points; s is (any) support vector
alphai is dual variables
K(.,.) is kernel
x is classify point

Notes
- Matlab classify returns (alphai.*yi) after training.
- The second sum using *one* support vector in the formula can be calculated more accurately by replacing with the mean of the same quantity on *all* support vectors

Bruno