A Generalized K-Means Algorithm with Semi-Supervised Weight Coefficients

Authors:
Fujiki Morii
Affiliations:
Nara Women's University, Nara 630-8506, Japan
Venue:
ICPR '06 Proceedings of the 18th International Conference on Pattern Recognition - Volume 03
Year:
2006

Citing 0
Cited 3

Supporting ranking and clustering as generalized order-by and group-by

Proceedings of the 2007 ACM SIGMOD international conference on Management of data
Using trees to depict a forest

Proceedings of the VLDB Endowment
A clustering based approach for skyline diversity

Expert Systems with Applications: An International Journal

Quantified Score

Hi-index	0.00

Visualization

Abstract

A new classification algorithm corresponding to a generalization of the K-means algorithm is proposed, whose algorithm is named as a weighted K-means algorithm. Weight coefficients, which provide weighted distortions between data and cluster centers, are incorporated into the algorithm to realize reliable classification. A method determining the appropriate values of the weight coefficients from class labeled data is introduced. Under the situations where statistical distributions of data are changing gradually with time, the weighted K-means algorithm for semi-supervised data composed from initial labeled data and succeeding unlabeled data is investigated.