It has been empirically or theoretically shown that a better learning machine with high generalization performance can be obtained by combining outputs of multiple learning machines. This is called, ensemble learnin9, a practical framework for constructing predictors with high generalization ability. In this tutorial, first, I explain the basic idea of ensemble learning and introduce several representative ensemble learning methods. I also give some intuitive and theoretical reasons why ensemble learning can improve generalization performance in some cases.
展开▼