REDES NEURONALES CON APRENDIZAJE NO SUPERVISADO
4.3.5. Simulaciones en MATLAB
El ejemplo anterior puede ser simulado en MATLAB tecleando el siguiente código:
clear;echo on;clc;nntwarn off;
P= rands(3, 10);
%número de neuronas
S=9;
W= initsm(P,S)
M= nbdist(9,1);
tp= [50 800 1];
[W]= trainsm(W,M,P,tp)
echo off
La salida del programa fue:
![Cuadro de texto: P= rands(3, 10); %número de neuronas S=9; W= initsm(P,S) W = 0.0890 -0.0460 -0.0732 0.0890 -0.0460 -0.0732 0.0890 -0.0460 -0.0732 0.0890 -0.0460 -0.0732 0.0890 -0.0460 -0.0732 0.0890 -0.0460 -0.0732 0.0890 -0.0460 -0.0732 0.0890 -0.0460 -0.0732 0.0890 -0.0460 -0.0732 M= nbdist(9,1); tp= [50 800 1]; [W]= trainsm(W,M,P,tp) TRAINSM: 0/800 epochs, neighborhood = 8, lr = 1. TRAINSM: 50/800 epochs, neighborhood = 2.18102, lr = 0.391853. TRAINSM: 100/800 epochs, neighborhood = 1, lr = 0.26182. TRAINSM: 200/800 epochs, neighborhood = 1, lr = 0.124875. TRAINSM: 300/800 epochs, neighborhood = 1, lr = 0.0615712. TRAINSM: 400/800 epochs, neighborhood = 1, lr = 0.0314628. TRAINSM: 500/800 epochs, neighborhood = 1, lr = 0.0169384. TRAINSM: 600/800 epochs, neighborhood = 1, lr = 0.00981873. TRAINSM: 700/800 epochs, neighborhood = 1, lr = 0.00624705. TRAINSM: 800/800 epochs, neighborhood = 1, lr = 0.00439161. W = -0.0193 0.6667 -0.8727 0.2825 0.3835 -0.7513 0.1315 -0.2561 -0.6138 -0.3447 -0.6444 -0.3148 -0.0189 -0.7583 0.1976 0.5841 -0.6739 0.3511 0.3928 -0.1568 0.4000 -0.1120 0.5531 0.6401 -0.1119 0.7276 0.6691 echo off](4.3.5_image174.png)
Figura 4.22: Salida Ejemplo SOM
