Role of Different Fuzzy Min- Max Neural Network for Pattern Classification
Journal: International Journal of Science and Research (IJSR) (Vol.3, No. 12)Publication Date: 2014-12-05
Authors : Jaitra Chakraborti;
Page : 1343-1346
Keywords : Fuzzy minmax FMM model; hyperbox structure; neural network learning; online learning; pattern classification;
Abstract
Different neural networks related to Fuzzy min-max (FMM) has been studied and amongst all, Enhanced Fuzzy min-max (EFMM) neural network is most recent. For classification of patterns a new Enhanced Fuzzy Min-Max (EFMM) algorithm has been studied. The aim of EFMM is to improve the performance and minimize the restrictions that are possessed by original fuzzy min-max (FMM) network. Three heuristic rules are used to improve the learning algorithm of FMM. First, to eliminate the problem of overlapping during hyperbox expansion, new overlapping rules has been suggested. Second, to discover other overlapping cases the hyperbox test rule has been extended. Third, to resolve the hyperbox overlapping cases, hyperbox contraction rule is provided.
Other Latest Articles
- Kamdhenu Ark as a Genoprotective Agent Against Chlorpyrifos Induced Genotoxicit
- Two Fluid Cosmological Model in A Kantowski-Sachs Space-Time in General Relativity
- Experimental Investigation of R152a/R22 Mixture in an Air Conditioning System
- Quasi-Static Thermal Stresses in a Thin Rectangular Plate Due to Heat Generation
- Polyaniline Nanoparticles Prepared by Micro-Emulsion Polymerization for Corrosion Inhibition of Carbon Steel in 1M HCl
Last modified: 2021-06-30 21:15:01