Document Detail


EM in high-dimensional spaces.
MedLine Citation:
PMID:  15971925     Owner:  NLM     Status:  MEDLINE    
Abstract/OtherAbstract:
This paper considers fitting a mixture of Gaussians model to high-dimensional data in scenarios where there are fewer data samples than feature dimensions. Issues that arise when using principal component analysis (PCA) to represent Gaussian distributions inside Expectation-Maximization (EM) are addressed, and a practical algorithm results. Unlike other algorithms that have been proposed, this algorithm does not try to compress the data to fit low-dimensional models. Instead, it models Gaussian distributions in the (N - 1)-dimensional space spanned by the N data samples. We are able to show that this algorithm converges on data sets where low-dimensional techniques do not.
Authors:
Bruce A Draper; Daniel L Elliott; Jeremy Hayes; Kyungim Baek
Related Documents :
17338825 - An efficient grid layout algorithm for biological networks utilizing various biological...
2079035 - Numerical technique and computational procedure for isotachophoresis.
20649425 - Sensibility of linkage information and effectiveness of estimated distributions.
18270115 - Integrating color and shape-texture features for adaptive real-time object tracking.
25472665 - A meta-analysis of the relationship between mthfr gene a1298c polymorphism and the risk...
22262465 - The impact of bias in the crown rump length measurement on first trimester screening fo...
Publication Detail:
Type:  Comparative Study; Evaluation Studies; Letter    
Journal Detail:
Title:  IEEE transactions on systems, man, and cybernetics. Part B, Cybernetics : a publication of the IEEE Systems, Man, and Cybernetics Society     Volume:  35     ISSN:  1083-4419     ISO Abbreviation:  IEEE Trans Syst Man Cybern B Cybern     Publication Date:  2005 Jun 
Date Detail:
Created Date:  2005-06-23     Completed Date:  2005-07-19     Revised Date:  2006-11-15    
Medline Journal Info:
Nlm Unique ID:  9890044     Medline TA:  IEEE Trans Syst Man Cybern B Cybern     Country:  United States    
Other Details:
Languages:  eng     Pagination:  571-7     Citation Subset:  IM    
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms
Descriptor/Qualifier:
Algorithms*
Artificial Intelligence*
Cluster Analysis
Computer Simulation
Image Enhancement / methods
Image Interpretation, Computer-Assisted / methods*
Information Storage and Retrieval / methods*
Likelihood Functions
Models, Biological
Models, Statistical
Pattern Recognition, Automated / methods*
Principal Component Analysis
Reproducibility of Results
Sensitivity and Specificity
Signal Processing, Computer-Assisted
Subtraction Technique

From MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine


Previous Document:  Image transform bootstrapping and its applications to semantic scene classification.
Next Document:  Robust fusion of uncertain information.