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: Adaptive Manifold Clustering

Abstract: We extend the theoretical study of a recently proposed nonparametric clustering algorithm called Adaptive Weights Clustering (AWC). In particular, we are interested in the case of high-dimensional data lying in the vicinity of a lower-dimensional non-linear submanifold with positive reach. After a slight adjustment and under rather general assumptions for the cluster structure, the algorithm turns out to be nearly optimal in detecting local inhomogeneities, while aggregating homogeneous data with a high probability. We also adress the problem of parameter tuning.
Subjects: Statistics Theory (math.ST)
Cite as: arXiv:1912.04869 [math.ST]
  (or arXiv:1912.04869v1 [math.ST] for this version)

Submission history

From: Franz Besold [view email]
[v1] Tue, 10 Dec 2019 18:24:36 GMT (2053kb,D)

Link back to: arXiv, form interface, contact.