Please enable JavaScript.
Coggle requires JavaScript to display documents.
BOOSTING - Coggle Diagram
BOOSTING
History
- one of the most powerful ideas introduced in ML
- It was first introduced in 1997 by Freund and Schapire
What?
- key idea is to use multiple weak learners to create a strong learner
- other way is to have high error only on rare data points and put more focus on data points which are common
- series of models are added which focus more on data points which have been mis classified by earlier models
NOTE
- It is a type of ensemble method similar to Bagging
- Bagging techniques are used in RandomForests algorithms, whereas Boosting used in ADABOOST, GradientBoost
-