Improved Vapnik Cervonenkis bounds
Catoni, Olivier
HAL, hal-00003056 / Harvested from HAL
Text on HAL
We give a new proof of VC bounds where we avoid the use of symmetrization and use a shadow sample of arbitrary size. We also improve on the variance term. This results in better constants, as shown on numerical examples. Moreover our bounds still hold for non identically distributed independent random variables. Keywords: Statistical learning theory, PAC-Bayesian theorems, VC dimension.
Publié le : 2004-10-11
Classification:  PAC-Bayesian theorems,  VC dimension,  Statistical learning theory,  62H30, 68T05, 62B10,  [MATH.MATH-PR]Mathematics [math]/Probability [math.PR] 
@article{hal-00003056,
     author = {Catoni, Olivier},
     title = {Improved Vapnik Cervonenkis bounds},
     journal = {HAL},
     volume = {2004},
     number = {0},
     year = {2004},
     language = {en},
     url = {http://dml.mathdoc.fr/item/hal-00003056}
}

Catoni, Olivier. Improved Vapnik Cervonenkis bounds. HAL, Tome 2004 (2004) no. 0, . http://gdmltest.u-ga.fr/item/hal-00003056/