Material Detail

α-Clusterable Sets

α-Clusterable Sets

This video was recorded at European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD), Athens 2011. In spite of the increasing interest into clustering research within the last decades, a unified clustering theory that is independent of a particular algorithm, or underlying the data structure and even the objective function has not be formulated so far. In the paper at hand, we take the first steps towards a theoretical foundation of clustering, by proposing a new notion of "clusterability" of data sets based on the density of the data within a specific region. Specifically, we give a formal definition of what we call "α-clusterable" set and we utilize this notion to prove that the principles proposed in Kleinberg's impossibility theorem for clustering [25], are consistent. We further propose an unsupervised clustering algorithm which is based on the notion of α-clusterable set. The proposed algorithm exploits the ability of the well known and widely used particle swarm optimization [31] to maximize the recently proposed window density function [38]. The obtained clustering quality is compared favorably to the corresponding clustering quality of various other well-known clustering algorithms.

Quality

  • User Rating
  • Comments
  • Learning Exercises
  • Bookmark Collections
  • Course ePortfolios
  • Accessibility Info

More about this material

Comments

Log in to participate in the discussions or sign up if you are not already a MERLOT member.