arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.03814stat.MLcs.ITcs.LGmath.ITmath.PR
收藏

Rényi散度的变分表示与神经网络估计

Variational Representations and Neural Network Estimation of Rényi Divergences

University of Massachusetts Amherst(马萨诸塞大学阿默斯特分校);Brown University(布朗大学)

Jeremiah Birrell, Paul Dupuis, Markos A. Katsoulakis, Luc Rey-Bellet, Jie Wang

摘要:

我们导出了概率测度$Q$和$P$之间的Rényi族分歧$R_α(Q\|P)$的一个新的变分公式。我们的结果推广了经典的关于库尔贝克-莱布勒散度的Donsker-Varadhan变分公式.我们进一步表明,这个雷尼变分公式在一系列函数空间上成立;这导致在非常弱的假设下的优化公式,也是我们发展雷尼发散估计的一致性理论的关键。通过将这一理论应用于神经网络估计,我们证明了如果一个神经网络族满足几个加强版本的泛逼近性质之一,那么相应的Rényi散度估计是一致的。与基于密度估计的方法相比,我们的估计只涉及$Q$和$P$下的期望,因此在高维系统中更有效。我们说明了这一点,通过几个数值的例子,神经网络估计系统的尺寸高达5000。

英文摘要:

We derive a new variational formula for the Rényi family of divergences, $R_α(Q\|P)$, between probability measures $Q$ and $P$. Our result generalizes the classical Donsker-Varadhan variational formula for the Kullback-Leibler divergence. We further show that this Rényi variational formula holds over a range of function spaces; this leads to a formula for the optimizer under very weak assumptions and is also key in our development of a consistency theory for Rényi divergence estimators. By applying this theory to neural-network estimators, we show that if a neural network family satisfies one of several strengthened versions of the universal approximation property then the corresponding Rényi divergence estimator is consistent. In contrast to density-estimator based methods, our estimators involve only expectations under $Q$ and $P$ and hence are more effective in high dimensional systems. We illustrate this via several numerical examples of neural network estimation in systems of up to 5000 dimensions.

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。

↑