arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2005.11818cs.LGcs.DSstat.ML
收藏

正确学习、Helly数和最优SVM界

Proper Learning, Helly Number, and an Optimal SVM Bound

Google Research, Brain Team(谷歌研究院大脑团队);Toyota Technological Institute at Chicago(芝加哥丰田理工学院)

Olivier Bousquet, Steve Hanneke, Shay Moran, Nikita Zhivotovskiy

摘要:

经典的PAC样本复杂性界限是针对任何经验风险最小化(ERM)的,并包含一个额外的对数因子$\log(1/ε)$,这是已知的是必要的一般ERM。Hanneke(2016)最近已经证明,任何VC类C的PAC学习的最佳样本复杂度是通过特定的不适当学习算法实现的,该算法输出C中特定的多数假设投票。这就留下了一个问题,即何时可以通过适当的学习算法来实现这个界限,这些算法被限制为总是从C输出一个假设。在本文中,我们的目标是表征类的最佳样本复杂度可以通过适当的学习算法。我们确定,这些类的特点是对偶Helly数,这是一个组合参数,出现在离散几何和抽象凸。特别是,在一般条件下的C,我们证明了对偶Helly数是有界的,当且仅当有一个适当的学习器,获得最佳的联合依赖于$ε$和$δ$。作为我们的技术的进一步影响,我们解决了Vapnik和Chervonenkis(1974)提出的一个长期悬而未决的问题,支持向量机的性能证明,在可实现的情况下,支持向量机的样本复杂性是$Θ((n/ε)+(1/ε)\log(1/δ))$,其中$n$是维度。通过适当的学习算法,给出了半空间的第一个最优PAC界,而且计算效率很高。

英文摘要:

The classical PAC sample complexity bounds are stated for any Empirical Risk Minimizer (ERM) and contain an extra logarithmic factor $\log(1/ε)$ which is known to be necessary for ERM in general. It has been recently shown by Hanneke (2016) that the optimal sample complexity of PAC learning for any VC class C is achieved by a particular improper learning algorithm, which outputs a specific majority-vote of hypotheses in C. This leaves the question of when this bound can be achieved by proper learning algorithms, which are restricted to always output a hypothesis from C. In this paper we aim to characterize the classes for which the optimal sample complexity can be achieved by a proper learning algorithm. We identify that these classes can be characterized by the dual Helly number, which is a combinatorial parameter that arises in discrete geometry and abstract convexity. In particular, under general conditions on C, we show that the dual Helly number is bounded if and only if there is a proper learner that obtains the optimal joint dependence on $ε$ and $δ$. As further implications of our techniques we resolve a long-standing open problem posed by Vapnik and Chervonenkis (1974) on the performance of the Support Vector Machine by proving that the sample complexity of SVM in the realizable case is $Θ((n/ε)+(1/ε)\log(1/δ))$, where $n$ is the dimension. This gives the first optimal PAC bound for Halfspaces achieved by a proper learning algorithm, and moreover is computationally efficient.

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。

↑