Training Classifiers under Covariate Shift by Constructing the Maximum Consistent Distribution Subset

Algorithm 5

The indirect classification algorithm of Scenario 2.

Require: the target training set , the target testing set

 , a base learning algorithm Learner

Ensure: classification function

 ;     / denote the feature vectors set of /

 ;  /constructing the MIDS/

 ;  /add the corresponding labels to the feature

            vectors of /

 ;  / denotes the function implemented by a base

        learning algorithm trained on the new training set /