A Dynamic Merge-or-Split Learning Algorithm on Gaussian Mixture for Automated Model Selection

Ma, Jinwen; He, Qi-Cai

doi:10.1007/11508069_27

Cited by 9 publications

(6 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…There are several widely used classes of methods for inferring the number of clusters or Gaussian components: such as measure-based [3] [4][5] [6], Bayesian [7] [8] and populationbased [9] [10], to name a few. The problem of inferring the number of components is a special case of the model selection problem and measure-based class of methods is the most widely used.…”

Section: Introductionmentioning

confidence: 99%

Learning the number of gaussian components using hypothesis test

Heo

Gader

2009

2009 International Joint Conference on Neural Networks

View full text Add to dashboard Cite

This paper addresses the problem of estimating the correct number of components in a Gaussian mixture given a sample data set. In particular, an extension of Gaussianmeans (G-means) and Projected Gaussian-means (PG-means) algorithms is proposed. All these methods are based on onedimensional statistical hypothesis test. G-means and PG-means are wrapper algorithms of the k-means and ExpectationMaximization (EM) algorithms, respectively. Although G-means is a simple and fast algorithm, it does not perform well when clusters overlap since it is based on k-means. PG-means can handle overlapped clusters but requires more computation and sometimes fails to find the right number of clusters.In this paper, we propose an extension, called Extended Projected Gaussian means (XPG-means). XPG-means is a wrapper algorithm of Possibilistic Fuzzy C-means (PFCM) algorithm. XPG-means integrates the advantages of both algorithms while resolving some of the disadvantages involving overlapped clusters, noise, and computational complexity. More specifically, XPG-means handles overlapped clusters better than G-means because of the use of fuzzy clustering, handles noise better than both algorithms because it uses possibilitistic clustering. XPG-means is less computationally expensive than PG-means because it uses local hypothesis testing scheme used by G-means that is specific to Gaussians wherease PG-means uses a more general Kolmogorov-Smirnow test on Gaussian mixtures. In addition, XPG-means demonstrates less variance in estimating the number of components than either of the other algorithms.

show abstract

Section: Introductionmentioning

confidence: 99%

Learning the number of gaussian components using hypothesis test

Heo

Gader

2009

2009 International Joint Conference on Neural Networks

View full text Add to dashboard Cite

show abstract

“…In [3] the authors developed a new methodology for fully Bayesian mixture analysis, using reversible jump Markov chain Monte Carlo methods to jumping the parameter subspaces and the different numbers of components in the mixture, while [4] propose a new kind of dynamic merge-or-split learning (DMOSL) algorithm to deal with the selection of number of Gaussians in the mixture, [5] describe an EM algorithm for nonparametric maximum likelihood (ML) estimation with variance component structure, [6] introduces a greedy algorithm for learning Gaussian mixture model, using combination of global and local search, [7] introduced a splitand-merge operation in order to alleviate the problem of local convergence of the usual EM algorithm, in paper [8] the authors use split-merge algorithm to analyze the data from the tracked video data.…”

Section: Introductionmentioning

confidence: 99%

Split-merge algorithm and Gaussian mixture models for AAL

Yin

Bruckner

2010

2010 IEEE International Symposium on Industrial Electronics

View full text Add to dashboard Cite

Analyzing time series sensor data and build statistical model in real time has to overcome two problems at least: the data count increase with time and the distribution of the data is dynamically. To deal with this kind of problems Gaussian mixture model and split-merge algorithm provide useful way.In an AAL project we handle the time series sensor data from a medical box contactor and a meal entrance contactor. Using Gaussian mixture model and split-merge algorithm to analyze the sensor data gathered for about one and a half months and built the statistical model.

show abstract

“…There are many resources about Gaussian models and fast learning algorithms: The authors of [1] develop iterative learning algorithms for estimating a coefficient vector in a given setting of statistical models with unknown hyper parameters, [2] described the globally supervised algorithm for Gaussian mixture models based on the maximum relative entropy (MRE), [3] is about an iterative algorithm for entropy regularized likelihood (ERL) learning on Gaussian mixture, [4] proposed a new kind of dynamic merge-or-split learning (DMOSL) algorithm on Gaussian mixture such that the number of Gaussians can be determined automatically. [5] presents a fast approximation method, based on kdtrees, reduces both the prediction and the training times of Gaussian process regression, [6] proposed a fast fixed-point learning algorithm for efficiently implementing maximization of the harmony function on Gaussian mixture with automated model selection.…”

Section: Introductionmentioning

confidence: 99%

Gaussian models and fast learning algorithm for persistence analysis of tracked video objects

Yin

Bruckner

2009

2009 2nd Conference on Human System Interactions

View full text Add to dashboard Cite

-Persistence of objects in scenes is an important parameter of video object tracking systems. From the analysis of objects' durations (of stay) we not only get how long they stay in the scene, but also precisely where the objects spend time. The video frame is therefore segmented into clusters, and objects which go through or stay there are assigned to that cluster. If we observe all objects in a time period we should get a model of object behavior with respect to duration for each cluster. Using the built model we try to find abnormal object behavior. To build a model of object's spatial duration from the video data we utilize Gaussians and fast learning algorithm for real time surveillance applications on embedded systems.

show abstract

A Dynamic Merge-or-Split Learning Algorithm on Gaussian Mixture for Automated Model Selection

Cited by 9 publications

References 10 publications

Learning the number of gaussian components using hypothesis test

Learning the number of gaussian components using hypothesis test

Split-merge algorithm and Gaussian mixture models for AAL

Gaussian models and fast learning algorithm for persistence analysis of tracked video objects

Contact Info

Product

Resources

About