AdaBoost

Creator
Creator
Seonglae Cho
Created
Created
2021 Oct 6 14:39
Editor
Edited
Edited
2025 Mar 24 22:0
Refs
Refs

AdaBoost modifies the data at each iteration

Adaboost modifies the data to weighted data at each iteration. At each step, the samples that were misclassified by previous predictor receive a higher weight to increase attention to previously misclassified samples. Thus, the classifier tries harder to classify them correctly. Weights for correctly classified samples are decreased
underfitted됐던 학습 데이터 샘플의 가중치를 높이면서 새로 학습된 모델이 학습하기 어려운 데이터에 더 잘 적합되도록 하는 방식이다.
notion image
여기선 멀리있는놈들이 과소적합weighted linear combination
가중치 선형결합을 하여 h(x) strong learner를 구하여 값을 예측한다.
 
 
 
 
 
 
 

Recommendations