Testing mutual independence in high dimensions with sums of squares of rank correlations
By Dennis Leung, Mathias Drton. 2015
9 citations recorded by the source corpus.
We treat the problem of testing mutual independence between m continuous observations when the available sample size n is comparable to m. As test statistics, we consider sums of squared rank correlations between pairs of random variables. Specific examples we study are sums of squares formed from Kendall's tau and from Spearman's rho. In the asymptotic regime where m/n converges to a positive constant, the null distribution of these statistics is shown to converge to a normal limit. The proofs are based on results for sums of squares of rank-based U-statistics.
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
- Bernhard Schölkopf
- Jan Peters
- Niklas Pfister
- Peter Buhlmann
- Yoshua Bengio
- Sergey Levine
- Xianyang Zhang
- David Lopez-Paz
Citations in the corpus
- Kernel‐based tests for joint independence (2016, background): “, 2005, 2007), and other related approaches for estimating or testing pairwise (in)dependence including distance correlations (Székely and Rizzo, 2009, 2014), rank-based correlations (Bergsma and Dassios, 2014; Leung and Drton, 2016; Nandy et al., 2016) or also non-parametric and semiparametric copu”
- Testing mutual independence in high dimension via distance covariance (2016, methodology): “The only exception is Bergsma and Dassios’s (2014) tÅtest, which was further extended by Leung and Drton (2017) to high dimension; some simulation comparisons between our distance-covariance-based test and the tÅ-based counterpart are provided.”