CLUSTERING USING SIMULATED ANNEALING WITH PROBABILISTIC REDISTRIBUTION

Article Properties
  • Language
    English
  • Publication Date
    2001/03/01
  • Indian UGC (Journal)
  • Refrences
    12
  • SANGHAMITRA BANDYOPADHYAY Machine Intelligence Unit, Indian Statistical Institute, 203 B.T. Road, Calcutta 700 035, India
  • UJJWAL MAULIK Department of Computer Science and Technology, Kalyani Government Engineering College, Kalyani 741 235, India
  • MALAY KUMAR PAKHIRA Department of Computer Science and Technology, Kalyani Government Engineering College, Kalyani 741 235, India
Abstract
Cite
BANDYOPADHYAY, SANGHAMITRA, et al. “CLUSTERING USING SIMULATED ANNEALING WITH PROBABILISTIC REDISTRIBUTION”. International Journal of Pattern Recognition and Artificial Intelligence, vol. 15, no. 02, 2001, pp. 269-85, https://doi.org/10.1142/s0218001401000927.
BANDYOPADHYAY, S., MAULIK, U., & PAKHIRA, M. K. (2001). CLUSTERING USING SIMULATED ANNEALING WITH PROBABILISTIC REDISTRIBUTION. International Journal of Pattern Recognition and Artificial Intelligence, 15(02), 269-285. https://doi.org/10.1142/s0218001401000927
BANDYOPADHYAY S, MAULIK U, PAKHIRA MK. CLUSTERING USING SIMULATED ANNEALING WITH PROBABILISTIC REDISTRIBUTION. International Journal of Pattern Recognition and Artificial Intelligence. 2001;15(02):269-85.
Journal Categories
Science
Mathematics
Instruments and machines
Electronic computers
Computer science
Technology
Electrical engineering
Electronics
Nuclear engineering
Electronics
Technology
Mechanical engineering and machinery
Description

Can a new clustering technique improve accuracy? This research introduces SAKM-clustering, an efficient partitional clustering technique that merges simulated annealing for minimum energy configuration with the K-means algorithm's searching capability. The method searches for appropriate clusters in multidimensional feature space to optimize a similarity metric. Data points are probabilistically redistributed, giving points farther from the cluster center a higher chance of migrating. The algorithm has been demonstrated for artificial and real life datasets and is compared to the K-means algorithm. The algorithm utilizes the power of simulated annealing for energy configuration and the searching capability of K-means. The method is particularly useful in large data sets that are multidimensional and complex.

This paper, published in the International Journal of Pattern Recognition and Artificial Intelligence, fits squarely within the journal's scope by presenting a novel clustering algorithm and demonstrating its performance on various datasets. The integration of simulated annealing aligns with the journal's interest in advanced computational techniques.

Refrences