Hi,
I am using LIBSVM software with MATLAB interface to classify my data which use SVM algorithm. It is giving output for decision values but I want distance from the hyperplane. In FAQ suggestions are given to find distance:
http://www.csie.ntu.edu.tw/~cjlin/libsvm/faq.html#f415
I am from mathematics background. So, it is difficult for me to modify the code according to the suggestion.
Online modifications are given to find distance using Python.
https://github.com/VanessaD/libsvm-3.0.extended.wdi.
But, I don't know python.
Can any one help me to modify the code.
Thanks.
Hi, if you are more familiar with MATLAB I suggest to modify the plotboundary function here explained
http://openclassroom.stanford.edu/MainFolder/DocumentPage.php?course=MachineLearning&doc=exercises/ex8/ex8.html
Once you have the model, you can use the "plotboundary" command that we have provided to visualize the decision boundary. The plotboundary command includes an option for seeing the filled-in contours of the function. Remember that this function gives the decision values that are used to make a classification.
Gino & Kirill,
Thanks a lot!!
Gino your suggestion is good but may be difficult for me to implement.
Kirill, I have done some modification in svm.cpp. I hope it is correct. I am attaching the code which i have done in svm.cpp.
I think I am correctly calculating the |w|.
Thanks again.
- Badges
- Science topic
- Similar topics
- Mathematics