Variants Extra Trees (Extremely Randomized Trees) More randomness: splits are random, not optimal Faster training Sometimes better performance Balanced Random Forest Handles imbalanced data better Each tree trained on balanced bootstrap sample