Hands-On Artificial Intelligence for Cybersecurity

Boosting algorithms

The boosting method, on the other hand, uses weighed samples that have been extracted from the data, whose weights are readjusted iteratively based on the classification errors that have been reported by the individual classifiers to reduce their bias.

Greater importance (weight) is given to the most difficult classification observations.

One of the best-known boosting algorithms is Adaptive Boosting (AdaBoost), in which a first classifier is trained on the training set.

The weight associated with the samples that are incorrectly classified by the first classifier is then incremented, a second classifier is trained on the dataset containing the updated weights, and so on. The iterative process ends when the predetermined ...