Abstract
In this paper, we discuss dimension reduction methods for 2D visualization of high dimensional clustered data. We propose a two-stage framework for visualizing such data based on dimension reduction methods. In the first stage, we obtain the reduced dimensional data by applying a supervised dimension reduction method such as linear discriminant analysis which preserves the original cluster structure in terms of its criteria. The resulting optimal reduced dimension depends on the optimization criteria and is often larger than 2. In the second stage, the dimension is further reduced to 2 for visualization purposes by another dimension reduction method such as principal component analysis. The role of the second-stage is to minimize the loss of information due to reducing the dimension all the way to 2. Using this framework, we propose several two-stage methods, and present their theoretical characteristics as well as experimental comparisons on both artificial and real-world text data sets.
| Original language | English |
|---|---|
| Title of host publication | VAST 09 - IEEE Symposium on Visual Analytics Science and Technology, Proceedings |
| Pages | 67-74 |
| Number of pages | 8 |
| DOIs | |
| Publication status | Published - 2009 Dec 1 |
| Externally published | Yes |
| Event | VAST 09 - IEEE Symposium on Visual Analytics Science and Technology - Atlantic City, NJ, United States Duration: 2009 Oct 12 → 2009 Oct 13 |
Publication series
| Name | VAST 09 - IEEE Symposium on Visual Analytics Science and Technology, Proceedings |
|---|
Conference
| Conference | VAST 09 - IEEE Symposium on Visual Analytics Science and Technology |
|---|---|
| Country | United States |
| City | Atlantic City, NJ |
| Period | 09/10/12 → 09/10/13 |
Fingerprint
Keywords
- 2D projection
- Clustered data
- Dimension reduction
- Generalized singular value decomposition
- Linear discriminant analysis
- Orthogonal centroid method
- Principal component analysis
- Regularization
ASJC Scopus subject areas
- Artificial Intelligence
- Computer Vision and Pattern Recognition
- Information Systems
Cite this
Two-stage framework for visualization of clustered high dimensional data. / Choo, Jaegul; Bohn, Shawn; Park, Haesun.
VAST 09 - IEEE Symposium on Visual Analytics Science and Technology, Proceedings. 2009. p. 67-74 5332629 (VAST 09 - IEEE Symposium on Visual Analytics Science and Technology, Proceedings).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Two-stage framework for visualization of clustered high dimensional data
AU - Choo, Jaegul
AU - Bohn, Shawn
AU - Park, Haesun
PY - 2009/12/1
Y1 - 2009/12/1
N2 - In this paper, we discuss dimension reduction methods for 2D visualization of high dimensional clustered data. We propose a two-stage framework for visualizing such data based on dimension reduction methods. In the first stage, we obtain the reduced dimensional data by applying a supervised dimension reduction method such as linear discriminant analysis which preserves the original cluster structure in terms of its criteria. The resulting optimal reduced dimension depends on the optimization criteria and is often larger than 2. In the second stage, the dimension is further reduced to 2 for visualization purposes by another dimension reduction method such as principal component analysis. The role of the second-stage is to minimize the loss of information due to reducing the dimension all the way to 2. Using this framework, we propose several two-stage methods, and present their theoretical characteristics as well as experimental comparisons on both artificial and real-world text data sets.
AB - In this paper, we discuss dimension reduction methods for 2D visualization of high dimensional clustered data. We propose a two-stage framework for visualizing such data based on dimension reduction methods. In the first stage, we obtain the reduced dimensional data by applying a supervised dimension reduction method such as linear discriminant analysis which preserves the original cluster structure in terms of its criteria. The resulting optimal reduced dimension depends on the optimization criteria and is often larger than 2. In the second stage, the dimension is further reduced to 2 for visualization purposes by another dimension reduction method such as principal component analysis. The role of the second-stage is to minimize the loss of information due to reducing the dimension all the way to 2. Using this framework, we propose several two-stage methods, and present their theoretical characteristics as well as experimental comparisons on both artificial and real-world text data sets.
KW - 2D projection
KW - Clustered data
KW - Dimension reduction
KW - Generalized singular value decomposition
KW - Linear discriminant analysis
KW - Orthogonal centroid method
KW - Principal component analysis
KW - Regularization
UR - http://www.scopus.com/inward/record.url?scp=72849129962&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=72849129962&partnerID=8YFLogxK
U2 - 10.1109/VAST.2009.5332629
DO - 10.1109/VAST.2009.5332629
M3 - Conference contribution
AN - SCOPUS:72849129962
SN - 9781424452835
T3 - VAST 09 - IEEE Symposium on Visual Analytics Science and Technology, Proceedings
SP - 67
EP - 74
BT - VAST 09 - IEEE Symposium on Visual Analytics Science and Technology, Proceedings
ER -