Abstract:We present a study of a kernel-based two-sample test statistic related to the Maximum Mean Discrepancy (MMD) in the manifold data setting, assuming that high-dimensional observations are close to a low-dimensional manifold. We characterize the test level and power in relation to the kernel bandwidth, the number of samples, and the intrinsic dimensionality of the manifold. Specifically, when data densities $p$ and $q$ are supported on a $d$-dimensional sub-manifold ${M}$ embedded in an $m$-dimensional space and are Hölder with order $\beta$ (up to 2) on ${M}$, we prove a guarantee of the test power for finite sample size $n$ that exceeds a threshold depending on $d$, $\beta$, and $\Delta_2$ the squared $L^2$-divergence between $p$ and $q$ on the manifold, and with a properly chosen kernel bandwidth $\gamma$. For small density departures, we show that with large $n$ they can be detected by the kernel test when $\Delta_2$ is greater than $n^{- { 2 \beta/( d + 4 \beta ) }}$ up to a certain constant and $\gamma$ scales as $n^{-1/(d+4\beta)}$. The analysis extends to cases where the manifold has a boundary and the data samples contain high-dimensional additive noise. Our results indicate that the kernel two-sample test has no curse-of-dimensionality when the data lie on or near a low-dimensional manifold. We validate our theory and the properties of the kernel test for manifold data through a series of numerical experiments.

Proof Of A Conjecture On Spearman'S P And Kendall'S Tau For Sample Minimum And Maximum

ON THE COMPLETE CONVERGENCE OF U-STATISTICS

A Supplement to the Complete Convergence

Upper Bound for the Expected Minimum of Dependent Random Variables with Known Kendall's Tau

On the Non-degeneracy of Kendall's and Spearman's Correlation Coefficients

A Limit Result for Self-Normalized Random Sums

A link between Kendall's tau, the length measure and the surface of bivariate copulas, and a consequence to copulas with self-similar support

A sharp inequality for Kendall's $τ$ and Spearman's $ρ$ of Extreme-Value Copulas

The Convergence of the Sums of Independent Random Variables under the Sub-Linear Expectations

On the minimum information checkerboard copulas under fixed Kendall's rank correlation

Maximal inequalities and a law of the iterated logarithm for negatively associated random fields

Strong Approximation Theorems for Sums of Random Variables when Extreme Terms Are Excluded

Dynamics Of The Bounds Of Squared Concurrence (Vol 79, 032306, 2009)

On kernel-based estimation of conditional Kendall's tau: finite-distance bounds and asymptotic behavior

Maximum interpoint distance of high-dimensional random vectors

On The Minimum Sum Rate Of Gaussian Multiterminal Source Coding: New Proofs

A Proof of the Convergence Theorem of Maximum-Entropy Clustering Algorithm

Kendall's Tau for Functional Data Analysis

The discrepancy in min-max statistics between two random matrices with finite third moments

Improved performance guarantees for Tukey's median

Kernel Two-Sample Tests for Manifold Data