Local Minima Structures in Gaussian Mixture Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Yudong, Song, Dogyoon, Xi, Xumei, Zhang, Yuqian
Format: Preprint
Published: 2020
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911380666318848
author Chen, Yudong
Song, Dogyoon
Xi, Xumei
Zhang, Yuqian
author_facet Chen, Yudong
Song, Dogyoon
Xi, Xumei
Zhang, Yuqian
contents We investigate the landscape of the negative log-likelihood function of Gaussian Mixture Models (GMMs) with a general number of components in the population limit. As the objective function is non-convex, there can be multiple local minima that are not globally optimal, even for well-separated mixture models. Our study reveals that all local minima share a common structure that partially identifies the cluster centers (i.e., means of the Gaussian components) of the true location mixture. Specifically, each local minimum can be represented as a non-overlapping combination of two types of sub-configurations: fitting a single mean estimate to multiple Gaussian components or fitting multiple estimates to a single true component. These results apply to settings where the true mixture components satisfy a certain separation condition, and are valid even when the number of components is over- or under-specified. We also present a more fine-grained analysis for the setting of one-dimensional GMMs with three components, which provide sharper approximation error bounds with improved dependence on the separation.
format Preprint
id arxiv_https___arxiv_org_abs_2009_13040
institution arXiv
publishDate 2020
record_format arxiv
spellingShingle Local Minima Structures in Gaussian Mixture Models
Chen, Yudong
Song, Dogyoon
Xi, Xumei
Zhang, Yuqian
Machine Learning
Statistics Theory
We investigate the landscape of the negative log-likelihood function of Gaussian Mixture Models (GMMs) with a general number of components in the population limit. As the objective function is non-convex, there can be multiple local minima that are not globally optimal, even for well-separated mixture models. Our study reveals that all local minima share a common structure that partially identifies the cluster centers (i.e., means of the Gaussian components) of the true location mixture. Specifically, each local minimum can be represented as a non-overlapping combination of two types of sub-configurations: fitting a single mean estimate to multiple Gaussian components or fitting multiple estimates to a single true component. These results apply to settings where the true mixture components satisfy a certain separation condition, and are valid even when the number of components is over- or under-specified. We also present a more fine-grained analysis for the setting of one-dimensional GMMs with three components, which provide sharper approximation error bounds with improved dependence on the separation.
title Local Minima Structures in Gaussian Mixture Models
topic Machine Learning
Statistics Theory
url https://arxiv.org/abs/2009.13040