arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2002.01013math.STstat.TH
收藏

高维光滑全变差的极限分布与$χ^2$ -发散

Limit Distribution for Smooth Total Variation and $χ^2$-Divergence in High Dimensions

Ziv Goldfeld, Kengo Kato

摘要:

统计偏差在机器学习中无处不在,作为测量概率分布之间差异的工具。由于这些应用程序固有地依赖于近似分布的样本,我们认为经验近似下两个流行的$f$ -分歧:全变差(TV)的距离和$χ^2$ -分歧。为了避免这些分歧的敏感性,以支持失配,高斯平滑的框架。研究了$\sqrt{n}δ_{\mathsf{TV}}(P_n\ast\mathcal{N},P\ast\mathcal{N})$和$nχ^2(P_n\ast\mathcal{N}\|P\ast\mathcal{N})$的极限分布,其中$P_n$是基于$n$独立同分布(i.i.d.)来自$P$、$\mathcal{N}_σ:=\mathcal{N}(0,σ^2\mathrm{I}_d)$和$\ast$的观测代表卷积。在任意维数下,极限分布可用$\mathbb{R}^d$上的高斯过程来刻画,协方差算子依赖于$P$和参数$σ$的各向同性高斯密度.这反过来又意味着最近为$δ_{\mathsf{TV}}(P_n\ast\mathcal{N},P\ast\mathcal{N})$和$χ^2(P_n\ast\mathcal{N}\|P\ast\mathcal{N})$导出的$n^{-1/2}$预期值收敛率的最优性。这些强有力的统计保证促进了高斯平滑下的经验近似,作为基于高维数据的学习和推理的有效框架。

英文摘要:

Statistical divergences are ubiquitous in machine learning as tools for measuring discrepancy between probability distributions. As these applications inherently rely on approximating distributions from samples, we consider empirical approximation under two popular $f$-divergences: the total variation (TV) distance and the $χ^2$-divergence. To circumvent the sensitivity of these divergences to support mismatch, the framework of Gaussian smoothing is adopted. We study the limit distributions of $\sqrt{n}δ_{\mathsf{TV}}(P_n\ast\mathcal{N},P\ast\mathcal{N})$ and $nχ^2(P_n\ast\mathcal{N}\|P\ast\mathcal{N})$, where $P_n$ is the empirical measure based on $n$ independently and identically distributed (i.i.d.) observations from $P$, $\mathcal{N}_σ:=\mathcal{N}(0,σ^2\mathrm{I}_d)$, and $\ast$ stands for convolution. In arbitrary dimension, the limit distributions are characterized in terms of Gaussian process on $\mathbb{R}^d$ with covariance operator that depends on $P$ and the isotropic Gaussian density of parameter $σ$. This, in turn, implies optimality of the $n^{-1/2}$ expected value convergence rates recently derived for $δ_{\mathsf{TV}}(P_n\ast\mathcal{N},P\ast\mathcal{N})$ and $χ^2(P_n\ast\mathcal{N}\|P\ast\mathcal{N})$. These strong statistical guarantees promote empirical approximation under Gaussian smoothing as a potent framework for learning and inference based on high-dimensional data.

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。

↑