We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.ST

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Statistics Theory

Title: Kernel-based Tests for Joint Independence

Abstract: We investigate the problem of testing whether $d$ random variables, which may or may not be continuous, are jointly (or mutually) independent. Our method builds on ideas of the two variable Hilbert-Schmidt independence criterion (HSIC) but allows for an arbitrary number of variables. We embed the $d$-dimensional joint distribution and the product of the marginals into a reproducing kernel Hilbert space and define the $d$-variable Hilbert-Schmidt independence criterion (dHSIC) as the squared distance between the embeddings. In the population case, the value of dHSIC is zero if and only if the $d$ variables are jointly independent, as long as the kernel is characteristic. Based on an empirical estimate of dHSIC, we define three different non-parametric hypothesis tests: a permutation test, a bootstrap test and a test based on a Gamma approximation. We prove that the permutation test achieves the significance level and that the bootstrap test achieves pointwise asymptotic significance level as well as pointwise asymptotic consistency (i.e., it is able to detect any type of fixed dependence in the large sample limit). The Gamma approximation does not come with these guarantees; however, it is computationally very fast and for small $d$, it performs well in practice. Finally, we apply the test to a problem in causal discovery.
Comments: 67 pages
Subjects: Statistics Theory (math.ST); Machine Learning (stat.ML)
Cite as: arXiv:1603.00285 [math.ST]
  (or arXiv:1603.00285v3 [math.ST] for this version)

Submission history

From: Niklas Pfister [view email]
[v1] Tue, 1 Mar 2016 14:34:53 GMT (83kb,D)
[v2] Sat, 30 Apr 2016 12:06:39 GMT (100kb)
[v3] Fri, 4 Nov 2016 10:20:59 GMT (916kb)

Link back to: arXiv, form interface, contact.