ARTMAP_Learn: ARTMAP_Learn

In gbaquer3/fuzzyARTMAP: Implementation of a fuzzy ARTMAP neural network

Description

Trains an ARTMAP network on the given input data.

Usage

ARTMAP_Learn(artmap_network, data, supervisor)

Arguments

artmap_network	ARTMAP network to be trained It should be created with ARTMAP_Create_Network().
data	Training data to be presented to the network. It is a matrix of size NumFeatures-by-NumSamples.
supervisor	the correct classification for each input vector. It is a matrix of size 1-by-NumSamples.

Details

This function trains an ARTMAP network on the given input data. Each sample of the data is presented to the network, which categorizes each sample and compares that category's entry in the map field to the supervisor signal. If the map field value and the supervisor signal do not match, match-tracking is induced until a category is found to code the input correctly. This category then learns the input vector.

The function returns a new ARTMAP network which has learned the input data according to the supervisor signal. If the maximum number of categories is reached and an appropriate categorization of the input cannot be made, no learning occurs. The program prints out a warning message that the maximum category limit has been reached and begins to process the next input vector.

Value

New ARTMAP network which has learned the input data.

gbaquer3/fuzzyARTMAP documentation built on Aug. 2, 2020, 6:33 p.m.