Output Modulation MethodIt is a method of transforming output representations and causing perturbations. It is often used to increase the diversity of learners. It can transform classification outputs into regression outputs and then construct individual learners.

The idea of perturbation of output representation is to manipulate the output representation through some operations to increase diversity.

Output disturbance

In addition to the output modulation method, there are several other ways to achieve output representation disturbance:

• Flip method: randomly change the labels of some training samples;
• ECOC method: Use error-correcting output codes to decompose a multi-classification task into a series of binary classification tasks, and use them to train the base learner. This is a method of decomposing the original task into multiple subtasks that can be solved simultaneously.
• Diversity Enhancement