Guardado en:
Detalles Bibliográficos
Autores principales: Dunlap, Alexander, Mourrat, Jean-Christophe
Formato: Preprint
Publicado: 2021
Materias:
Acceso en línea:https://arxiv.org/abs/2104.13753
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866913349311135744
author Dunlap, Alexander
Mourrat, Jean-Christophe
author_facet Dunlap, Alexander
Mourrat, Jean-Christophe
contents Sum-of-norms clustering is a popular convexification of $K$-means clustering. We show that, if the dataset is made of a large number of independent random variables distributed according to the uniform measure on the union of two disjoint balls of unit radius, and if the balls are sufficiently close to one another, then sum-of-norms clustering will typically fail to recover the decomposition of the dataset into two clusters. As the dimension tends to infinity, this happens even when the distance between the centers of the two balls is taken to be as large as $2\sqrt{2}$. In order to show this, we introduce and analyze a continuous version of sum-of-norms clustering, where the dataset is replaced by a general measure. In particular, we state and prove a local-global characterization of the clustering that seems to be new even in the case of discrete datapoints.
format Preprint
id arxiv_https___arxiv_org_abs_2104_13753
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle Sum-of-norms clustering does not separate nearby balls
Dunlap, Alexander
Mourrat, Jean-Christophe
Machine Learning
Statistics Theory
Sum-of-norms clustering is a popular convexification of $K$-means clustering. We show that, if the dataset is made of a large number of independent random variables distributed according to the uniform measure on the union of two disjoint balls of unit radius, and if the balls are sufficiently close to one another, then sum-of-norms clustering will typically fail to recover the decomposition of the dataset into two clusters. As the dimension tends to infinity, this happens even when the distance between the centers of the two balls is taken to be as large as $2\sqrt{2}$. In order to show this, we introduce and analyze a continuous version of sum-of-norms clustering, where the dataset is replaced by a general measure. In particular, we state and prove a local-global characterization of the clustering that seems to be new even in the case of discrete datapoints.
title Sum-of-norms clustering does not separate nearby balls
topic Machine Learning
Statistics Theory
url https://arxiv.org/abs/2104.13753