Maximizing the Margin with Boosting
AdaBoost produces a linear combination of weak hypotheses. It has been observed that the generalization error of the algorithm continues to improve even after all examples are classified correctly by the current linear combination, i.e. by a hyperplane in feature space spanned by the weak hypotheses. The improvement is attributed to the experimental observation that the distances (margins) of the examples to the separating hyperplane are increasing even when the training error is already zero, that is all examples are on the correct side of the hyperplane. We give an iterative version of AdaBoost that explicitly maximizes the minimum margin of the examples. We bound the number of iterations and the number of hypotheses used in the final linear combination which approximates the maximum margin hyperplane with a certain precision. Our modified algorithm essentially retains the exponential convergence properties of AdaBoost and our result does not depend on the size of the hypothesis class.
KeywordsBase Learner Training Error Generalization Error Maximum Margin Base Hypothesis
Unable to display preview. Download preview PDF.
- 1.K.P. Bennett, A. Demiriz, and J. Shawe-Taylor. A column generation algorithm for boosting. In P. Langley, editor, Proceedings, 17th ICML, pages 65–72, San Francisco, 2000.Google Scholar
- 3.Y. Freund. Boosting a weak learning algorithm by majority. Information and Computation, 121(2):256–285, September 1995.Google Scholar
- 4.Y. Freund and R.E. Schapire. Experiments with a new boosting algorithm. In Proc. 13th International Conference on Machine Learning, pages 148–146. Morgan Kaufmann, 1996.Google Scholar
- 7.A.J. Grove and D. Schuurmans. Boosting in the limit: Maximizing the margin of learned ensembles. In Proc. of the Fifteenth National Conference on Artifical Intelligence, 1998.Google Scholar
- 8.R. Hettich and K.O. Kortanek. Semi-infinite programming: Theory, methods and applications. SIAM Review, 3:380–429, September 1993.Google Scholar
- 9.J. Kivinen and M. Warmuth. Boosting as entropy projection. In Proc. 12th Annu. Conference on Comput. Learning Theory, pages 134–144. ACM Press, New York, NY, 1999.Google Scholar
- 10.V. Koltchinskii, D. Panchenko, and F. Lozano. Some new bounds on the generalization error of combined classifiers. In Advances in Neural Inf. Proc. Systems, volume 13, 2001.Google Scholar
- 12.S. Nash and A. Sofer. Linear and Nonlinear Programming. McGraw-Hill, New York, 1996.Google Scholar
- 13.J.R. Quinlan. C4.5: Programs for Machine Learning. Morgan Kaufmann, 1992.Google Scholar
- 14.J.R. Quinlan. Boosting first-order learning. Lecture Notes in Comp. Sci., 1160:143, 1996.Google Scholar
- 15.G. Rätsch. Robust Boosting via Convex Optimization. PhD thesis, University of Potsdam, October 2001. http://mlg.anu.edu.au/~raetsch/thesis.ps.gz.
- 16.G. Rätsch, A. Demiriz, and K. Bennett. Sparse regression ensembles in infinite and finite hypothesis spaces. Machine Learning, 48(1–3):193–221, 2002. Special Issue on New Methods for Model Selection and Model Combination. Also NeuroCOLT2 Technical Report 2000-085.Google Scholar
- 17.G. Rätsch, T. Onoda, and K.-R. Müller. Soft margins for AdaBoost. Machine Learning, 42(3):287–320, March 2001. also NeuroCOLT Technical Report NC-TR-1998-021.Google Scholar
- 18.R.E. Schapire. The Design and Analysis of Efficient Learning Algorithms. PhD thesis, MIT Press, 1992.Google Scholar
- 20.R.E. Schapire and Y. Singer. Improved boosting algorithms using confidence-rated predictions. Machine Learning, 37(3):297–336, December 1999. also Proceedings of the 14th Workshop on Computational Learning Theory 1998, pages 80–91.Google Scholar
- 23.T. Zhang. Sequential greedy approximation for certain convex optimization problems. Technical report, IBM T.J. Watson Research Center, 2002.Google Scholar