TY - JOUR
T1 - Probabilistic principal component subspaces
T2 - a hierarchical finite mixture model for data visualization
AU - Wang, Yue
AU - Luo, Lan
AU - Freedman, Matthew T.
AU - Kung, Sun Yuan
N1 - Funding Information:
Manuscript received September 18, 1999; revised February 28, 2000. This work was supported in part by the National Institutes of Health under Grant 1R21CA83231-01 and the Department of Defense under Grants DAMD17-98-1-8045 and DAMD17-98-1-8044. Y. Wang and L. Luo are with the Department of Electrical Engineering and Computer Science, The Catholic University of America, Washington, DC 20064 USA (e-mail: [email protected]). M. T. Freedman is with the Department of Radiology and the Lombardi Cancer Center, Georgetown University, Washington, DC 20007 USA. S. Y. Kung is with the Department of Electrical Engineering, Princeton University, Princeton, NJ 08544 USA (e-mail: [email protected]). Publisher Item Identifier S 1045-9227(00)04040-6.
PY - 2000/5
Y1 - 2000/5
N2 - Visual exploration has proven to be a powerful tool for multivariate data mining and knowledge discovery. Most visualization algorithms aim to find a projection from the data space down to a visually perceivable rendering space. To reveal all of the interesting aspects of multimodal data sets living in a high-dimensional space, a hierarchical visualization algorithm is introduced which allows the complete data set to be visualized at the top level, with clusters and subclusters of data points visualized at deeper levels. The methods involve hierarchical use of standard finite normal mixtures and probabilistic principal component projections, whose parameters are estimated using the expectation-maximization and principal component neural networks under the information theoretic criteria. We demonstrate the principle of the approach on several multimodal numerical data sets, and we then apply the method to the visual explanation in computer-aided diagnosis for breast cancer detection from digital mammograms.
AB - Visual exploration has proven to be a powerful tool for multivariate data mining and knowledge discovery. Most visualization algorithms aim to find a projection from the data space down to a visually perceivable rendering space. To reveal all of the interesting aspects of multimodal data sets living in a high-dimensional space, a hierarchical visualization algorithm is introduced which allows the complete data set to be visualized at the top level, with clusters and subclusters of data points visualized at deeper levels. The methods involve hierarchical use of standard finite normal mixtures and probabilistic principal component projections, whose parameters are estimated using the expectation-maximization and principal component neural networks under the information theoretic criteria. We demonstrate the principle of the approach on several multimodal numerical data sets, and we then apply the method to the visual explanation in computer-aided diagnosis for breast cancer detection from digital mammograms.
UR - http://www.scopus.com/inward/record.url?scp=0034187075&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0034187075&partnerID=8YFLogxK
U2 - 10.1109/72.846734
DO - 10.1109/72.846734
M3 - Article
C2 - 18249790
AN - SCOPUS:0034187075
SN - 1045-9227
VL - 11
SP - 625
EP - 636
JO - IEEE Transactions on Neural Networks
JF - IEEE Transactions on Neural Networks
IS - 3
ER -