-
Your decision boundary might be overtrained. Which means that if your training set is not including some examples that you want to have in a class, when you use those examples after training, you might not get the correct class label.
-
When this an input which is not from any of the classes in reality, then it might get a wrong class label after classification.
-
You have to select a lot of good examples from each class while you are training the classifier. If you consider classification of big data that can be a real challenge.
-
Training needs a lot of computation time, so do the classification.
-
You might need to use a cloud and leave the training algorithm work over a night or nights before obtaining a good decision boundary model.