Deterministic Annealing: A Variant of Simulated Annealing and its Application to Fuzzy Clustering
Author:
Open Access
- 26 April 2017
- book chapter
- Published by IntechOpen
Abstract
Deterministic annealing (DA) is a deterministic variant of simulated annealing. In this chapter, after briefly introducing DA, we explain how DA is combined with the fuzzy c-means (FCM) clustering by employing the entropy maximization method, especially the Tsallis entropy maximization. The Tsallis entropy is a q parameter extension of the Shannon entropy. Then, we focus on Tsallis-entropy-maximized FCM (Tsallis-DAFCM), and examine effects of cooling functions for DA on accuracy and convergence. A shape of a membership function of Tsallis-DAFCM depends on both a system temperature and q. Accordingly, a relationship between the temperature and q is quantitatively investigated.Keywords
This publication has 31 references indexed in Scilit:
- Statistical Mechanical Analysis of Fuzzy Clustering Based on Fuzzy EntropyIEICE Transactions on Information and Systems, 2007
- Possibilistic and probabilistic fuzzy clustering: unification within the framework of the non-extensive thermostatisticsPattern Recognition, 2003
- Deterministic annealing for clustering, compression, classification, regression, and related optimization problemsProceedings of the IEEE, 1998
- Pairwise data clustering by deterministic annealingIEEE Transactions on Pattern Analysis and Machine Intelligence, 1997
- A global optimization technique for statistical classifier designIEEE Transactions on Signal Processing, 1996
- Vector quantization with complexity costsIEEE Transactions on Information Theory, 1993
- Constrained clustering as an optimization methodIEEE Transactions on Pattern Analysis and Machine Intelligence, 1993
- Vector quantization by deterministic annealingIEEE Transactions on Information Theory, 1992
- Optimization by Simulated AnnealingScience, 1983
- An Algorithm for Vector Quantizer DesignIEEE Transactions on Communications, 1980