Additive logistic regression: A statistical view of boosting: Discussion

We congratulate Friedman, Hastie and Tibshirani (FHT) for their significant work connecting boosting and (a kind of) additive models. FHT provide a much-needed bridge between an important and effective machine learning procedure and traditional statistical modeling ideas. With this bridge, ideas can now flow easily in both directions so that a deeper and more thorough understanding of boosting will eventually emerge. In this discussion, we would like to share our thoughts and reflections on boosting and related statistical ideas, inspired by FHT.