What is the role of margin theory in Random Forest generalization?

Margin theory explains generalization based on confidence difference between correct and incorrect class votes.

The margin is defined as the difference between the proportion of correct class votes and the highest incorrect class votes. A larger margin implies better generalization bounds. Random Forest improves margin distribution by aggregating many weakly correlated classifiers, reducing probability of small-margin predictions and thus lowering generalization error bounds.