Localized Rademacher Complexities

  • Peter L. Bartlett
  • Olivier Bousquet
  • Shahar Mendelson
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 2375)


We investigate the behaviour of global and local Rademacher averages. We present new error bounds which are based on the local averages and indicate how data-dependent local averages can be estimated without a priori knowledge of the class at hand.


Error Bound Generalization Error Entire Class Empirical Risk Class Member 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    P. L. Bartlett, S. Boucheron and G. Lugosi. Model selection and error estimation. Machine Learning, 48, 85–113, 2002.zbMATHCrossRefGoogle Scholar
  2. 2.
    P. L. Bartlett and S. Mendelson. Rademacher and Gaussian Complexities: Risk Bounds and Structural Results. Proceedings of the Fourteenth Annual Conference on Computational Learning Theory, pp. 224–240, 2001.Google Scholar
  3. 3.
    S. Boucheron, G. Lugosi and P. Massart. Concentration inequalities using the entropy method. Preprint, CNRS-Université Paris-Sud. 2002.Google Scholar
  4. 4.
    O. Bousquet. A Bennett concentration inequality and its application to empirical processes. C. R. Acad. Sci. Paris, Ser. I, 334, pp. 495–500, 2002.zbMATHMathSciNetGoogle Scholar
  5. 5.
    V. I. Koltchinskii. Rademacher penalties and structural risk minimization. Technical report, University of New Mexico, 2000.Google Scholar
  6. 6.
    V. I. Koltchinskii and D. Panchenko. Rademacher processes and bounding the risk of function learning. In High Dimensional Probability II, Eds. E. Gine, D. Mason and J. Wellner, pp. 443–459, 2000.Google Scholar
  7. 7.
    W. S. Lee, P. L. Bartlett and R. C. Williamson. Efficient agnostic learning of neural networks with bounded fan-in. IEEE Transactions on Information Theory, 42(6), 2118–2132, 1996.zbMATHCrossRefMathSciNetGoogle Scholar
  8. 8.
    P. Massart. Some applications of concentration inequalities to statistics. Annales de la Faculté des Sciences de Toulouse, IX:245–303, 2000.MathSciNetGoogle Scholar
  9. 9.
    S. Mendelson. Improving the sample complexity using global data. IEEE Transactions on Information Theory, 2002.Google Scholar
  10. 10.
    S. Mendelson and R. C. Williamson Agnostic learning nonconvex classes of function. To appear in Proceedings of the Fifteenth Annual Conference on Computational Learning Theory, 2002.Google Scholar
  11. 11.
    E. Rio Une inegalité de Bennett pour les maxima de processus empiriques. Colloque en l’honneur de J. Bretagnolle, D. Dacunha-Castelle et I. Ibragimov, 2001.Google Scholar
  12. 12.
    A. W. van der Vaart and J. A. Wellner. Weak Convergence and Empirical Processes. Springer, 1996.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2002

Authors and Affiliations

  • Peter L. Bartlett
    • 1
    • 3
  • Olivier Bousquet
    • 1
    • 2
  • Shahar Mendelson
    • 3
  1. 1.BIOwulf TechnologiesBerkeleyUSA
  2. 2.Centre de Mathématiques AppliquéesEcole PolytechniquePalaiseauFrance
  3. 3.Research School of Information Sciences and EngineeringThe Australian National UniversityCanberraAustralia

Personalised recommendations