This chapter explores the principles and practical applications of the (Adaptive Boosting) AdaBoost algorithm, a leading technique in the family of boosting methods. Boosting operates by combining multiple weak learners—classifiers that perform only marginally better than random guessing—into a single, highly accurate ensemble. AdaBoost achieves this by iteratively adjusting the weights of training examples: examples that are misclassified by the current weak learner are assigned higher weights, while correctly classified examples receive lower weights. This adaptive weighting mechanism ensures that subsequent learners focus on the more challenging cases, thereby progressively improving the overall classifier’s performance. The chapter highlights AdaBoost’s strong theoretical guarantees, noting that if each weak learner performs better than random guess, the training error of the combined classifier decreases exponentially with the number of iterations. This rapid convergence underscores the algorithm’s efficiency. The generalization ability of AdaBoost is examined through the lens of margin theory, which relates a classifier’s margin—the difference between the confidence of the correct class and that of the most competitive incorrect class—to its likelihood of performing well on unseen data. The discussion extends to various AdaBoost adaptations, including AdaBoost.M2, which is tailored for multi-class classification tasks.