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: Score matching estimators for directional distributions

Abstract: One of the major problems for maximum likelihood estimation in the well-established directional models is that the normalising constants can be difficult to evaluate. A new general method of "score matching estimation" is presented here on a compact oriented Riemannian manifold. Important applications include von Mises-Fisher, Bingham and joint models on the sphere and related spaces. The estimator is consistent and asymptotically normally distributed under mild regularity conditions. Further, it is easy to compute as a solution of a linear set of equations and requires no knowledge of the normalizing constant. Several examples are given, both analytic and numerical, to demonstrate its good performance.
Comments: 21 pages, 2 figures, 5 tables
Subjects: Statistics Theory (math.ST)
MSC classes: 62F10 (Primary), 62H11 (Secondary)
Cite as: arXiv:1604.08470 [math.ST]
  (or arXiv:1604.08470v1 [math.ST] for this version)

Submission history

From: John Kent [view email]
[v1] Thu, 28 Apr 2016 15:36:43 GMT (30kb,D)

Link back to: arXiv, form interface, contact.