TITLE

Bayesian Inference for Nonnegative Matrix Factorisation Models

AUTHOR(S)
Cemgil, Ali Taylan
PUB. DATE
January 2009
SOURCE
Computational Intelligence & Neuroscience;2009, Special section p1
SOURCE TYPE
Academic Journal
DOC. TYPE
Article
ABSTRACT
We describe nonnegative matrix factorisation (NMF) with a Kullback-Leibler (KL) error measure in a statistical framework, with a hierarchical generative model consisting of an observation and a prior component. Omitting the prior leads to the standard KL-NMF algorithms as special cases, where maximum likelihood parameter estimation is carried out via the Expectation- Maximisation (EM) algorithm. Starting from this view, we develop full Bayesian inference via variational Bayes or Monte Carlo. Our construction retains conjugacy and enables us to develop more powerful models while retaining attractive features of standard NMF such as monotonic convergence and easy implementation. We illustrate our approach on model order selection and image reconstruction.
ACCESSION #
47715145

 

Related Articles

  • Combining standardized mean differences using the method of maximum likelihood. KE-HAI YUAN; BUSHMAN, BRAD J. // Psychometrika;Dec2002, Vol. 67 Issue 4, p589 

    A maximum likelihood procedure for combining standardized mean differences based on a noncentrat t-distribution is proposed. With a proper data augmentation technique, an EM-algorithm is developed. Information and likelihood ratio statistics are discussed in detail for reliable inference....

  • Estimation Methods for a Two-Parameter Bathtub-Shaped Lifetime Distribution. Mazucheli, Josmar; Ghitany, M. E.; Louzada, Francisco // Australian Journal of Basic & Applied Sciences;Jul2014, Vol. 8 Issue 10, p189 

    In this paper we compare several methods for estimating the parameters of a twoparameter lifetime distribution that can be used for modeling bathtub hazard rate behavior. The estimation methods considered are: maximum likelihood, ordinary least squares, weighted least squares, Anderson-Darling,...

  • Modified Moment, Maximum Likelihood and Percentile Estimators for the Parameters of the Power Function Distribution. Zaka, Azam; Akhter, Ahmad Saeed // Pakistan Journal of Statistics & Operation Research;2014, Vol. 10 Issue 4, p369 

    This paper is concerned with the modifications of maximum likelihood, moments and percentile estimators of the two parameter Power function distribution. Sampling behavior of the estimators is indicated by Monte Carlo simulation. For some combinations of parameter values, some of the modified...

  • Shrinkage Estimation of the Exponential mean With Grouped Data. Baklizi, Ayman // Annual International Conference on Operations Research & Statist;2016, p191 

    The problem of estimating the mean of the exponential distribution using grouped censored data is considered. Assuming that there is a prior information about the value of the mean, we developed two shrinkage estimators that utilize this prior information. The first suggested estimator is...

  • Alternative estimator for the parameters of a mixture of two binomial distributions. Quinino, Roberto; Ho, Linda; Suyama, Emílio // Statistical Papers;Feb2013, Vol. 54 Issue 1, p47 

    The method of moments has been widely used as a simple alternative to the maximum likelihood method, mainly because of its efficiency and simplicity in obtaining parameter estimators of a mixture of two binomial distributions. In this paper, an alternative estimate is proposed which is as...

  • Maximum Likelihood Estimator for Bearings-only Passive Target Tracking in Electronic Surveillance Measure and Electronic Warfare Systems. Rao, S. Koteswara // Defence Science Journal;2010, Vol. 60 Issue 2, p197 

    Maximum likelihood estimator is a suitable algorithm for passive target tracking applications. Nardone, Lindgren and Gong introduced this approach using batch processing. In this paper, the batch processing is converted into sequential processing for real-time applications like passive target...

  • Probabilistic Monte-Carlo Method for Modelling and Prediction of Electronics Component Life. Sreenuch, T.; Alghassi, A.; Perinpanayagam, S.; Xie, Y. // International Journal of Advanced Computer Science & Application;Jan2014, Vol. 5 Issue 1, p96 

    Power electronics are widely used in electric vehicles, railway locomotive and new generation aircrafts. Reliability of these components directly affect the reliability and performance of these vehicular platforms. In recent years, several research work about reliability, failure mode and aging...

  • A Note on Modified Estimators for the Parameters of the Power Function Distribution. Zaka, Azam; Feroze, Navid; Akhter, Ahmad Saeed // International Journal of Advanced Science & Technology;2013, Vol. 59, p71 

    This paper is concerned with the modifications of maximum likelihood, moments and percentile estimators of the two parameter Power function distribution. Sampling behavior of the estimators is indicated by Monte Carlo simulation. For some combinations of parameter values, some of the modified...

  • Particle Filter Algorithm for DOA Tracking of Maneuvering Targets. SONG De-shu; LIANG Guo-long; WANG Yan // Journal of Signal Processing;Jul2014, Vol. 30 Issue 7, p861 

    An improved particle filter(PF) algorithm is proposed to address the problems of tracking precision descend, bad real-time performance and large error against multiple targets tracking due to the direction-of-arrival(DOA) of maneuvering targets changing rapidly. According to the model of array...

Share

Read the Article

Courtesy of VIRGINIA BEACH PUBLIC LIBRARY AND SYSTEM

Sorry, but this item is not currently available from your library.

Try another library?
Sign out of this library

Other Topics