Efficient Bisecting k-Medoids and Its Application in Gene Expression Analysis

Authors:
Rasha Kashef;Mohamed S. Kamel
Affiliations:
Electrical and Computer Engineering Department, University of Waterloo,;Electrical and Computer Engineering Department, University of Waterloo,
Venue:
ICIAR '08 Proceedings of the 5th international conference on Image Analysis and Recognition
Year:
2008

Citing 6
Cited 1

Data clustering: a review

ACM Computing Surveys (CSUR)
A clustering algorithm based on graph connectivity

Information Processing Letters
Clustering Algorithms

Clustering Algorithms
Efficient and Effective Clustering Methods for Spatial Data Mining

VLDB '94 Proceedings of the 20th International Conference on Very Large Data Bases
Incorporating Gene Ontology in Clustering Gene Expression Data

CBMS '06 Proceedings of the 19th IEEE Symposium on Computer-Based Medical Systems
Validity-guided (re)clustering with applications to image segmentation

IEEE Transactions on Fuzzy Systems

Application of K-Medoids with Kd-Tree for Software Fault Prediction

ACM SIGSOFT Software Engineering Notes

Quantified Score

Hi-index	0.00

Visualization

Abstract

The medoid-based clustering algorithm, Partition Around Medoids (PAM), is better than the centroid-based k-means because of its robustness to noisy data and outliers. PAM cannot recognize relatively small clusters in situations where good partitions around medoids clearly exist. Also PAM needs O(k(n-k)2) operations to cluster a given dataset, which is computationally prohibited for large nand k. In this paper, we propose a new bisecting k-medoids algorithm that is capable of grouping the co-expressed genes together with better clustering quality and time performances. The proposed algorithm is evaluated over three gene expression datasets in which noise components are involved. The proposed algorithm takes less computation time with comparable performance relative to the Partitioning Around Medoids algorithm.