arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 0907.3740stat.ML
收藏

经验Bernstein界与样本方差惩罚

Empirical Bernstein Bounds and Sample Variance Penalization

Andreas Maurer, Massimiliano Pontil

摘要:

我们给出了数据依赖和方差敏感的置信界的改进常数,称为经验Bernstein界,并将这些不等式推广到增长函数在样本量n中为多项式的函数类上.界限导致我们考虑样本方差惩罚,一种新的学习方法,考虑到经验方差的损失函数。给出了样本方差惩罚有效的条件。特别是,我们提出了一个绑定的方法所产生的超额风险。利用这一点,我们认为,有情况下,我们的方法的超额风险是为了1/n,而经验风险最小化的超额风险是为了1/sqrt/{n}。我们给出了一些实验结果,证实了这一理论。最后,我们讨论了潜在的应用,我们的结果样本压缩方案。

英文摘要:

We give improved constants for data dependent and variance sensitive confidence bounds, called empirical Bernstein bounds, and extend these inequalities to hold uniformly over classes of functionswhose growth function is polynomial in the sample size n. The bounds lead us to consider sample variance penalization, a novel learning method which takes into account the empirical variance of the loss function. We give conditions under which sample variance penalization is effective. In particular, we present a bound on the excess risk incurred by the method. Using this, we argue that there are situations in which the excess risk of our method is of order 1/n, while the excess risk of empirical risk minimization is of order 1/sqrt/{n}. We show some experimental results, which confirm the theory. Finally, we discuss the potential application of our results to sample compression schemes.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑