I implemented the co-training to undergraduate and now need to implement an ensemble of ensemble after each iteration.
For example: In the 1st iteration, we will have a classifier and the data will be labeled only by this classifier. In the 2nd iteration, we will have an ensemble to classify, and so on.
Since the co-training divides the data into multiple classifier to make this classification, I want to label each independent of the vision given at the end and make a vote or to average between views to label.
I need ideas to make the best strategy for implementation. I'm using the WEKA and already implemented the co-training as stated in the beginning.
"excuse the errors, do not speak english".
code: http://pastebin.com/Xd8guMub
code: http://pastebin.com/FL8Y2j0c
comments of the code in portuguese-brazil
