Unbiased Approach to Determine The Initial Conditions for Clustering Techniques
Original Publication Date: 2004-Oct-26
Included in the Prior Art Database: 2004-Oct-26
The proposed technique is a non-biased approach to select the initial values for the cluster means used for cluster-based techniques such as k-means clustering. Not only the final solution is unique and more representative of what the actual objects (clusters) are but the speed of convergence is noticeably improved by this approach. An application of this technique to a Satellite image of the RDU airport is demonstrated.