1
$\begingroup$

Liblinear (http://www.csie.ntu.edu.tw/~cjlin/liblinear/) does not support for probability estimates.

Say I have three classes C1, C2 and C3. I want to learn the model paramters for each 'one vs rest' cases:

       C1 vs C2&C3, 
       c2 vs C1&C3 and
       C3 vs C1&C2

How can I do it?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

To make 'one-vs-rest' comparisons, simply relabel all the classes you're comparing against. So in your first comparison, C1 stays C1 and C2 and C3 are grouped together as some new class like C4.

Improve this answer
$\endgroup$
4
  • $\begingroup$ But then how can I compare the final results? (thats why I mentioned that liblinear does not support for probability estimates). I need the probability estimates to get the final classification results. $\endgroup$
    – user570593
    Mar 5, 2013 at 13:04
  • $\begingroup$ What classifier are you using in Liblinear? I assume not an SVM else you would be using LibSVM, or might follow the Liblinear team's directions $\endgroup$
    – learner
    Mar 5, 2013 at 14:32
  • $\begingroup$ I am using SVM classification using libLinear. The problem is LibSVM supports probability estimates but libLinear does not support (supports probability estimates only for regression problems) $\endgroup$
    – user570593
    Mar 5, 2013 at 15:26
  • 1
    $\begingroup$ My advice would be either to use LibSVM or to look for implementations of Lin's revision to Platt's solution or to follow the linked directions on the Liblinear website. $\endgroup$
    – learner
    Mar 5, 2013 at 18:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.