What is the connection between Random Forest and bias-corrected bagging estimators?

Random Forest can be viewed as an enhanced bagging estimator with additional feature randomness reducing correlation bias.

Standard bagging reduces variance but retains correlation between base learners. Random Forest introduces feature subsampling, which decorrelates trees further, effectively reducing the covariance term in ensemble error decomposition. This acts as an implicit bias-correction mechanism by preventing dominance of strong predictors across all trees.