jordwest / comp4702-case-study

COMP4702 Case Study

A Novel Kernel Method For Clustering

This report gives an overview of the author’s attempt to replicate the results of the proposed clustering method from the paper “A Novel Kernel Method For Clustering” [1]. As the original paper does not provide complete details on the implementation of the proposed clustering algorithm (hereby referred to as the new algorithm), some assumptions are made in an attempt to replicate the algorithm. In addition, the result of K-means clustering on the same data sets is attempted for comparison to the original paper. The other existing clustering algorithms (SOM, Neural Gas, Ng-Jordan) tested in the original paper are not investigated in this report.