An Efficient Smooth Boosting by Filtering(New Trends in Theory of Computation and Algorithm)
Kohei Hatano · Kyoto University Research Information Repository (Kyoto University) · 2006
Boosting is a general method to construct a highly accurate classifier by combining "weakly" accurate ones.Smooth boosting algorithms are variants of boosting methods which handle only smooth distributions on the data.They are proved to be noise-tolerant and can be used in the "boosting by filtering" scheme, which is suitable for learning over huge data.However, current smooth boosting algorithms have rooms for improvements: A non-smooth boosting algorithm, InfoBoost can perform more efficiently than typical boosting algorithmns by using an information-theoretic criterion for choosing hypotheses.In this paper, we propose a new smooth boosting algorithm with an information-theoretic criterion and we show that it inherits the advantages of two approaches, smooth boosting and InfoBoost.