Seonglae ChoHighly randomizes both feature selection and split points during node splitting
Standard Random Forest:
- Randomly selects subset of features
- Optimizes split points (using information gain, etc.)
Extra-Trees:
- Features are randomly selected
- Split points are also randomly chosen
- Best split among random candidates is selected
High computational efficiency. Theoretically shown to work essentially like random split-based kernel approximation.
Extremely randomized trees
Machine Learning - This paper proposes a new tree-based ensemble method for supervised classification and regression problems. It essentially consists of randomizing strongly both attribute and...
https://link.springer.com/article/10.1007/s10994-006-6226-1
