Adjusting initial weights for Adaboost learning
Kisang Kim, Hyung–Il Choi · 2017
The Adaboost extracts an optimal set of weak classifiers in stages. On each stage, it chooses the optimal classifier by minimizing the weighted error classification. It also reweights training data so that the next round would focus on data that are difficult to classify. The typical Adaboost algorithm assigns the same weight to each training datum on the first round of a training process. In this paper, we propose to assign different initial weights based on some statistical properties of involved features. In experimental results, we assess that the proposed method shows higher performance than the typical one.