A Graph-Theoretic Approach to Cluster Analysis.

Abstract

The problem of cluster analysis is formulated in the realm of graph theory. A graph-theoretic clustering technique is presented which makes use of the concepts of the connectivity of a graph, the network flow theory, the invariance principle and the k-nearest-neighbor threshold method of edge construction. Several major drawbacks of the existing clustering methods can be overcome by the proposed method. Computer experiments of the application of the method to artificial and real data are reported.

Document Details

Document Type
Technical Report
Publication Date
Jul 01, 1975
Accession Number
ADA015928

Entities

People

  • King Sun Fu
  • Zexun Chen

Organizations

  • Purdue University

Tags

DTIC Thesaurus Topics

  • Clustering
  • Computer Science
  • Computers
  • Construction
  • Graph Theory
  • Invariance
  • Mathematics

Fields of Study

  • Computer science

Readers

  • Neural Network Machine Learning.
  • Theoretical Analysis.