Uploaded March 2021 | Updated November 2026, 1 day ago
ERRATA: The output of the AdaBoost ensemble is not between -1 and 1, but between -∞ and ∞.
In this video we look at boosting: ensembling methods that update the models in the ensemble sequentially, in order to combat bias. We look in detail at Adaboost, which works by reweighting the data at every step, and at gradient boosting, which requires each model to predict the residuals of the ensemble so far.
slides: mlvu.github.io/lectures/52.Trees.annotated.pdf
lecturer: Peter Bloem
ERRATA: The output of the AdaBoost ensemble is not between -1 and 1, but between -∞ and ∞.
In this video we look at boosting: ensembling methods that update the models in the ensemble sequentially, in order to combat bias. We look in detail at Adaboost, which works by reweighting the data at every step, and at gradient boosting, which requires each model to predict the residuals of the ensemble so far.
slides: mlvu.github.io/lectures/52.Trees.annotated.pdf
lecturer: Peter Bloem










