One-Pass Boosting.
Z. Barutcuoglu and P. Long and R. Servedio.
21st Annual Conference in Neural Information Processing Systems (NIPS), 2007.


Abstract:

We analyze the performance of boosting algorithms that make only a single pass over a collection of base classifiers. Our results indicate that ``picky'' boosters, which abstain from using base classifiers whose advantage is too small, can perform effectively in this setting.

Postscript or pdf.


Back to main papers page