Abstract
Clustering is a fundamental task in data mining. We propose, implement, and evaluate several schemes that combine partitioning and hierarchical algorithms, specifically k-means and principal direction divisive partitioning (PDDP). Using available theory regarding the solution of the clustering indicator vector problem, we use 2means to induce partitionings around fixed or varying cut-points. 2-means is applied either on the data or over its projection on a one-dimensional subspace. These techniques are also extended to the case of PDDP(l), a multiway clustering algorithm generalizing PDDP. To handle data that do not lend themselves to linear separability, the algebraic framework is established for a kernel variant, KPDDP. Extensive experiments demonstrate the performance of the above methods and suggest that it is advantageous to steer PDDP using k-means. It is also shown that KPDDP can provide results of superior quality than kernel k-means.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.