Multiple Fuzzification Coefficients in a Fuzzy C-Means Clustering Algorithm

  • Unique Paper ID: 165738
  • Volume: 10
  • Issue: 1
  • PageNo: 1672-1691
  • Abstract:
  • Clustering is a well researched unsupervised machine learning technique with numerous real-world applications. Besides probabilistic or deterministic methods, fuzzy C-means clustering (FCM) is another popular method for clustering. Clustering efficiency has improved significantly since the FCM method was introduced. These enhancements concentrate on modifying the distance function between elements and the membership representation of the elements in the clusters, or on fuzzifying and defuzzifying methods. This paper suggests a novel fuzzy clustering algorithm that makes use of several fuzzification factors, which are chosen based on the properties of individual data samples. With a few adjustments, the suggested fuzzy clustering approach uses computation steps that are comparable to FCM. Convergence is guaranteed by deriving the formulas. Using numerous fuzzification coefficients instead of the one coefficient used in the original FCM method is the primary contribution of this approach. Experiments on a number of widely used datasets are then used to assess the new algorithm, and the findings indicate that it is more effective than both the original FCM and alternative clustering techniques.
email to a friend

Cite This Article

  • ISSN: 2349-6002
  • Volume: 10
  • Issue: 1
  • PageNo: 1672-1691

Multiple Fuzzification Coefficients in a Fuzzy C-Means Clustering Algorithm

Related Articles