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

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

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

Computer Science > Machine Learning

Title: Functorial Hierarchical Clustering with Overlaps

Abstract: This work draws inspiration from three important sources of research on dissimilarity-based clustering and intertwines those three threads into a consistent principled functorial theory of clustering. Those three are the overlapping clustering of Jardine and Sibson, the functorial approach of Carlsson and M\'{e}moli to partition-based clustering, and the Isbell/Dress school's study of injective envelopes. Carlsson and M\'{e}moli introduce the idea of viewing clustering methods as functors from a category of metric spaces to a category of clusters, with functoriality subsuming many desirable properties. Our first series of results extends their theory of functorial clustering schemes to methods that allow overlapping clusters in the spirit of Jardine and Sibson. This obviates some of the unpleasant effects of chaining that occur, for example with single-linkage clustering. We prove an equivalence between these general overlapping clustering functors and projections of weight spaces to what we term clustering domains, by focusing on the order structure determined by the morphisms. As a specific application of this machinery, we are able to prove that there are no functorial projections to cut metrics, or even to tree metrics. Finally, although we focus less on the construction of clustering methods (clustering domains) derived from injective envelopes, we lay out some preliminary results, that hopefully will give a feel for how the third leg of the stool comes into play.
Comments: Minor revisions. 24 pages, 1 figure
Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
MSC classes: 51K05, 68P01
ACM classes: H.3.3
Journal reference: Discrete Applied Mathematics, Volume 236, 19 February 2018, pp.108--123
DOI: 10.1016/j.dam.2017.10.015
Cite as: arXiv:1609.02513 [cs.LG]
  (or arXiv:1609.02513v2 [cs.LG] for this version)

Submission history

From: Dan Guralnik [view email]
[v1] Thu, 8 Sep 2016 17:52:26 GMT (167kb)
[v2] Tue, 14 Aug 2018 18:34:04 GMT (168kb)

Link back to: arXiv, form interface, contact.