Interval-Valued Centroids in K-Means Algorithms
Benjamine Nordin, Chenyi Hu, Bernard K. Chen, Victor S. Sheng · 2012
The K-Means algorithms are fundamental in machine learning and data mining. In this study, we investigate interval-valued rather than commonly used point-valued centroids in the K-Means algorithm. Using a proposed interval peak method to select initial interval centroids, we have obtained overall quality improvement of clusters on a set of test problems in the Fundamental Clustering Problem Suite (FCPS).