Fuzzy ART and Fuzzy ARTMAP Neural Networks
22 Dec 2003
24 Dec 2003)
This package allows creation, training, and testing of ART and ARTMAP neural networks.
% Fuzzy ART Neural Network Implementation.
% Version 1.0 02-Apr-2002
% Aaron Garrett
% Jacksonville State University
% Jacksonville, AL 36265
% Functions included in this directory:
% ART_Activate_Categories - Performs the network category activation for a given input.
% ART_Add_New_Category - Adds a new category element to the ART network.
% ART_Calculate_Match - Calculates the degree of match between a given input and a category.
% ART_Categorize - Uses a trained ART network to categorize a dataset.
% ART_Complement_Code - Complement-codes the given input.
% ART_Create_Network - Creates the ART network.
% ART_Learn - Trains a given ART network on a dataset.
% ART_Update_Weights - Updates the weight matrix of the network.
% Description of the system architecture:
% The above set of functions is used to create, train, and use an ART
% network to categorize a dataset. While all of the functions are
% necessary, only half of them are meant to be called by the user.
% Those functions are as follows:
% Functions available to user:
% The remaining four functions are used to modularize the structure
% of the system. These functions are related to different components
% of adaptive resonance theory.
% ART_Activate_Categories essentially provides bottom-up activation
% of the F2 layer for a given input.
% ART_Add_New_Category is used after a series of mismatch resets in
% order to create a new F2 neuron to code the current input.
% ART_Calculate_Match is used to determine the degree of match between
% a given input and the category coded by the current F2 neuron.
% ART_Update_Weights is used to update the weight matrix during learning
% after resonance has been achieved.
% For an example of the use of these functions, see ARTExample.m in
% this directory.
% Send comments to email@example.com.