Variational expectation-maximization training for Gaussian networks

N Nasios; A G Bors

Variational expectation-maximization training for Gaussian networks

Computer Science

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

Abstract

This paper introduces variational expectation-maximization (VEM) algorithm for training Gaussian networks. Hyperparameters model distributions of parameters characterizing Gaussian mixture densities. The proposed algorithm employs a hierarchical learning strategy for estimating a set of hyperparameters and the number of Gaussian mixture components. A dual EM algorithm is employed as the initialization stage in the VEM-based learning. In the first stage the EM algorithm is applied on the given data set while the second stage EM is used on distributions of parameters resulted from several runs of the first stage EM. Appropriate maximum log-likelihood estimators are considered for all the parameter distributions involved.

Original language	English
Title of host publication	2003 IEEE XIII WORKSHOP ON NEURAL NETWORKS FOR SIGNAL PROCESSING - NNSP'03
Place of Publication	NEW YORK
Publisher	IEEE
Pages	339-348
Number of pages	10
ISBN (Print)	0-7803-8177-7
Publication status	Published - 2003
Event	13th IEEE Workshop on Neural Networks for Signal Processing (NNSP 2003) - Toulouse Duration: 17 Sept 2003 → 19 Sept 2003

Conference

Conference	13th IEEE Workshop on Neural Networks for Signal Processing (NNSP 2003)
City	Toulouse
Period	17/09/03 → 19/09/03

Keywords

EM ALGORITHM
MODELS

Cite this

@inproceedings{0b20a0ba58ed4b04b50b2dedca215648,

title = "Variational expectation-maximization training for Gaussian networks",

abstract = "This paper introduces variational expectation-maximization (VEM) algorithm for training Gaussian networks. Hyperparameters model distributions of parameters characterizing Gaussian mixture densities. The proposed algorithm employs a hierarchical learning strategy for estimating a set of hyperparameters and the number of Gaussian mixture components. A dual EM algorithm is employed as the initialization stage in the VEM-based learning. In the first stage the EM algorithm is applied on the given data set while the second stage EM is used on distributions of parameters resulted from several runs of the first stage EM. Appropriate maximum log-likelihood estimators are considered for all the parameter distributions involved.",

keywords = "EM ALGORITHM, MODELS",

author = "N Nasios and Bors, {A G}",

year = "2003",

language = "English",

isbn = "0-7803-8177-7",

pages = "339--348",

booktitle = "2003 IEEE XIII WORKSHOP ON NEURAL NETWORKS FOR SIGNAL PROCESSING - NNSP'03",

publisher = "IEEE",

address = "United States",

note = "13th IEEE Workshop on Neural Networks for Signal Processing (NNSP 2003) ; Conference date: 17-09-2003 Through 19-09-2003",

}

TY - GEN

T1 - Variational expectation-maximization training for Gaussian networks

AU - Nasios, N

AU - Bors, A G

PY - 2003

Y1 - 2003

N2 - This paper introduces variational expectation-maximization (VEM) algorithm for training Gaussian networks. Hyperparameters model distributions of parameters characterizing Gaussian mixture densities. The proposed algorithm employs a hierarchical learning strategy for estimating a set of hyperparameters and the number of Gaussian mixture components. A dual EM algorithm is employed as the initialization stage in the VEM-based learning. In the first stage the EM algorithm is applied on the given data set while the second stage EM is used on distributions of parameters resulted from several runs of the first stage EM. Appropriate maximum log-likelihood estimators are considered for all the parameter distributions involved.

AB - This paper introduces variational expectation-maximization (VEM) algorithm for training Gaussian networks. Hyperparameters model distributions of parameters characterizing Gaussian mixture densities. The proposed algorithm employs a hierarchical learning strategy for estimating a set of hyperparameters and the number of Gaussian mixture components. A dual EM algorithm is employed as the initialization stage in the VEM-based learning. In the first stage the EM algorithm is applied on the given data set while the second stage EM is used on distributions of parameters resulted from several runs of the first stage EM. Appropriate maximum log-likelihood estimators are considered for all the parameter distributions involved.

KW - EM ALGORITHM

KW - MODELS

M3 - Conference contribution

SN - 0-7803-8177-7

SP - 339

EP - 348

BT - 2003 IEEE XIII WORKSHOP ON NEURAL NETWORKS FOR SIGNAL PROCESSING - NNSP'03

PB - IEEE

CY - NEW YORK

T2 - 13th IEEE Workshop on Neural Networks for Signal Processing (NNSP 2003)

Y2 - 17 September 2003 through 19 September 2003

ER -