Which boosting method weights misclassified data points more to focus learning on them?

Which boosting method weights misclassified data points more to focus learning on them?

• A. Boosting
• B. AdaBoost
• C. Random Forests
• D. K-Nearest Neighbors

Answer: (B)

Focusing learning on hard examples by reweighting misclassified data is a defining feature of AdaBoost. In this approach, you start with equal weights for all training points. After training a weak learner, you assess which points it got wrong. Those misclassified points have their weights increased for the next learner, so future models pay more attention to them. Points that were classified correctly may see their weights kept the same or reduced. Each weak learner also gets a weight based on its accuracy, and the final prediction is a weighted combination of all the learners' outputs. This targeted emphasis on difficult cases is what makes AdaBoost different from other methods listed. Random Forests build many trees independently using bootstrap samples and random feature subsets, without changing the importance of individual training cases across rounds. K-Nearest Neighbors makes predictions based on nearby labeled points in the feature space without iterating learners or adjusting weights. General boosting, while related, doesn’t specify this misclassification-weighting mechanism as explicitly as AdaBoost does.

Focusing learning on hard examples by reweighting misclassified data is a defining feature of AdaBoost. In this approach, you start with equal weights for all training points. After training a weak learner, you assess which points it got wrong. Those misclassified points have their weights increased for the next learner, so future models pay more attention to them. Points that were classified correctly may see their weights kept the same or reduced. Each weak learner also gets a weight based on its accuracy, and the final prediction is a weighted combination of all the learners' outputs.

This targeted emphasis on difficult cases is what makes AdaBoost different from other methods listed. Random Forests build many trees independently using bootstrap samples and random feature subsets, without changing the importance of individual training cases across rounds. K-Nearest Neighbors makes predictions based on nearby labeled points in the feature space without iterating learners or adjusting weights. General boosting, while related, doesn’t specify this misclassification-weighting mechanism as explicitly as AdaBoost does.