A Projected Clustering Algorithm and Its Biomedical Application - Clustering Challenges in Biological Network - page 186

Chapter 9
A Projected Clustering Algorithm and Its Biomedical Application
Ping Deng
Computer Science
University of Illinois at Springfield, USA
[e-mail]
Qingkai Ma
Economic Crime and Justice Studies
Utica College, USA
[e-mail]
Weili Wu ∗
Computer Science
The University of Texas at Dallas, USA
[e-mail]
Projected clustering is concerned with clustering data in high dimensional space
where data is more likely correlated in subspaces of full dimensions. Recently,
several projected clustering algorithms that focus on finding specific projection
for each cluster have been proposed. We find that, besides distance, the closeness
of points in different dimensions also depends on the distributions of data along
those dimensions. Based on this, we propose a projected clustering algorithm,
IPROCLUS (Improved PROCLUS), which is efficient and accurate in handling
data in high dimensional space. According to the experimental results on ran-
domly generated synthetic data, our algorithm shows much higher accuracy for
the scaled datasets and lower dependence on one of user inputs than PROCLUS.
We also apply IPROCLUS on real biomedical data and show that it can achieve
much better accuracy than PROCLUS.
∗ This work was partially supported by NSF grant # ACI-0305567. The content of this work does
not necessarily reflect the position or policy of the government and no official endorsement should be
inferred.
187