Clusterpath: An algorithm for clustering using convex fusion penalties

Toby Dylan Hocking; Armand Joulin; Francis Bach; Jean Philippe Vert

Clusterpath: An algorithm for clustering using convex fusion penalties

Toby Dylan Hocking, Armand Joulin, Francis Bach, Jean Philippe Vert

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We present a new clustering algorithm by proposing a convex relaxation of hierarchical clustering, which results in a family of objective functions with a natural geometric interpretation. We give efficient algorithms for calculating the continuous regularization path of solutions, and discuss relative advantages of the parameters. Our method experimentally gives state-of-the-art results similar to spectral clustering for non-convex clusters, and has the added benefit of learning a tree structure from the data.

Original language	English (US)
Title of host publication	Proceedings of the 28th International Conference on Machine Learning, ICML 2011
Pages	745-752
Number of pages	8
State	Published - 2011
Externally published	Yes
Event	28th International Conference on Machine Learning, ICML 2011 - Bellevue, WA, United States Duration: Jun 28 2011 → Jul 2 2011

Publication series

Name	Proceedings of the 28th International Conference on Machine Learning, ICML 2011

Conference

Conference	28th International Conference on Machine Learning, ICML 2011
Country/Territory	United States
City	Bellevue, WA
Period	6/28/11 → 7/2/11

ASJC Scopus subject areas

Computer Science Applications
Human-Computer Interaction
Education

Cite this

Clusterpath: An algorithm for clustering using convex fusion penalties. / Hocking, Toby Dylan; Joulin, Armand; Bach, Francis et al.
Proceedings of the 28th International Conference on Machine Learning, ICML 2011. 2011. p. 745-752 (Proceedings of the 28th International Conference on Machine Learning, ICML 2011).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Hocking, TD, Joulin, A, Bach, F & Vert, JP 2011, Clusterpath: An algorithm for clustering using convex fusion penalties. in Proceedings of the 28th International Conference on Machine Learning, ICML 2011. Proceedings of the 28th International Conference on Machine Learning, ICML 2011, pp. 745-752, 28th International Conference on Machine Learning, ICML 2011, Bellevue, WA, United States, 6/28/11.

@inproceedings{30d5dc28b4f34a7dbedf227f5eeec3cd,

title = "Clusterpath: An algorithm for clustering using convex fusion penalties",

abstract = "We present a new clustering algorithm by proposing a convex relaxation of hierarchical clustering, which results in a family of objective functions with a natural geometric interpretation. We give efficient algorithms for calculating the continuous regularization path of solutions, and discuss relative advantages of the parameters. Our method experimentally gives state-of-the-art results similar to spectral clustering for non-convex clusters, and has the added benefit of learning a tree structure from the data.",

author = "Hocking, {Toby Dylan} and Armand Joulin and Francis Bach and Vert, {Jean Philippe}",

year = "2011",

language = "English (US)",

isbn = "9781450306195",

series = "Proceedings of the 28th International Conference on Machine Learning, ICML 2011",

pages = "745--752",

booktitle = "Proceedings of the 28th International Conference on Machine Learning, ICML 2011",

note = "28th International Conference on Machine Learning, ICML 2011 ; Conference date: 28-06-2011 Through 02-07-2011",

}

TY - GEN

T1 - Clusterpath

T2 - 28th International Conference on Machine Learning, ICML 2011

AU - Hocking, Toby Dylan

AU - Joulin, Armand

AU - Bach, Francis

AU - Vert, Jean Philippe

PY - 2011

Y1 - 2011

N2 - We present a new clustering algorithm by proposing a convex relaxation of hierarchical clustering, which results in a family of objective functions with a natural geometric interpretation. We give efficient algorithms for calculating the continuous regularization path of solutions, and discuss relative advantages of the parameters. Our method experimentally gives state-of-the-art results similar to spectral clustering for non-convex clusters, and has the added benefit of learning a tree structure from the data.

AB - We present a new clustering algorithm by proposing a convex relaxation of hierarchical clustering, which results in a family of objective functions with a natural geometric interpretation. We give efficient algorithms for calculating the continuous regularization path of solutions, and discuss relative advantages of the parameters. Our method experimentally gives state-of-the-art results similar to spectral clustering for non-convex clusters, and has the added benefit of learning a tree structure from the data.

UR - http://www.scopus.com/inward/record.url?scp=80053439464&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80053439464&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:80053439464

SN - 9781450306195

T3 - Proceedings of the 28th International Conference on Machine Learning, ICML 2011

SP - 745

EP - 752

BT - Proceedings of the 28th International Conference on Machine Learning, ICML 2011

Y2 - 28 June 2011 through 2 July 2011

ER -

Clusterpath: An algorithm for clustering using convex fusion penalties

Abstract

Publication series

Conference

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this