Global and Componentwise Extrapolation for Accelerating Data Mining from Large Incomplete Data Sets with the EM Algorithm

Hsu, Chun‐Nan; Huang, Han-Shen; Yang, Bo-Hou

doi:10.1109/icdm.2006.77

Cited by 3 publications

(3 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the situations where there is the huge missing data sets and many parameters, EM algorithm will converge slowly. Therefore, Hsu et al [12] has proposed the triple jump extrapolation method to solve the performance of EM by reducing the number of iterative convergence. In additional to that, Em algorithm is low effective when there is not the class number pre-assigned appropriately.…”

Section: ) Expectation-maximization (Em)mentioning

confidence: 99%

A survey of data mining techniques for analyzing crime patterns

Thongsatapornwatana¹

2016

2016 Second Asian Conference on Defence Technology (ACDT)

View full text Add to dashboard Cite

In recent years the data mining is data analyzing techniques that used to analyze crime data previously stored from various sources to find patterns and trends in crimes. In additional, it can be applied to increase efficiency in solving the crimes faster and also can be applied to automatically notify the crimes. However, there are many data mining techniques. In order to increase efficiency of crime detection, it is necessary to select the data mining techniques suitably. This paper reviews the literatures on various data mining applications, especially applications that applied to solve the crimes. Survey also throws light on research gaps and challenges of crime data mining. In additional to that, this paper provides insight about the data mining for finding the patterns and trends in crime to be used appropriately and to be a help for beginners in the research of crime data mining.

show abstract

Section: ) Expectation-maximization (Em)mentioning

confidence: 99%

A survey of data mining techniques for analyzing crime patterns

Thongsatapornwatana¹

2016

2016 Second Asian Conference on Defence Technology (ACDT)

View full text Add to dashboard Cite

show abstract

“…This happens when there are high percentages of missing data for a Bayesian network model trained by EM (Hsu et al 2006) and when features are uncorrelated for training a CRF model (Huang et al 2009). …”

Section: Now Letmentioning

confidence: 99%

“…The estimates γ and γ i have been applied in the framework of Aitken's acceleration to speed up the convergence of the EM algorithm (Hesterberg 2005;Huang et al 2005;Hsu et al 2006), suggesting that both γ and γ i can be sufficiently accurate to allow for substantial speedup for the convergence of EM.…”

Section: Aitken's Accelerationmentioning

confidence: 99%

Periodic step-size adaptation in second-order gradient descent for single-pass on-line structured learning

et al. 2009

Self Cite

View full text Add to dashboard Cite

It has been established that the second-order stochastic gradient descent (SGD) method can potentially achieve generalization performance as well as empirical optimum in a single pass through the training examples. However, second-order SGD requires computing the inverse of the Hessian matrix of the loss function, which is prohibitively expensive for structured prediction problems that usually involve a very high dimensional feature space. This paper presents a new second-order SGD method, called Periodic Step-size Adaptation (PSA). PSA approximates the Jacobian matrix of the mapping function and explores a linear relation between the Jacobian and Hessian to approximate the Hessian, which is proved to be simpler and more effective than directly approximating Hessian in an on-line setting. We tested PSA on a wide variety of models and tasks, including large scale sequence labeling tasks using conditional random fields and large scale classification tasks using linear support vector machines and convolutional neural networks. Experimental results show that single-pass performance of PSA is always very close to empirical optimum.

show abstract