Perceptrón multicapa java

Solo disponible en BuenasTareas
  • Páginas : 3 (738 palabras )
  • Descarga(s) : 10
  • Publicado : 22 de julio de 2010
Leer documento completo
Vista previa del texto
Manual Técnico:
Perceptrón multicapa:

Por
Hugo

El presente manual, tiene por objetivo mostrar la estructura general del programa, para que así se puedan hacer las mejoras y cambiosposteriores de manera fácil y óptima.
Se cuentan con los siguientes métodos:
Class perceptron{
* declaración de variables
* public void init()
* public void llenaPesos()
* public voidobtenerSalidaRed()
* public void backPropagation()
}

1. Variables:
* int iteracion, rTabla;
rTatbla es el renglón de la tabla de patrones
iteración es el número de iteraciones que habrá* double entrenamiento [][]= {
{0.6,0.44444444,0.175,0.01},
{0.68,0.18333333,0.85,0.01},
{0.84,0.3,0.55,0.01},
{0.76,0.27777778,0.5,0.01},
{1,0.24444444,0.475,0.01},{0.8,0.13888889,0.9,0.01},
{0.72,0.19555556,0.7,0.01},
{0.9,0.18888889,0.75,0.01},
{0.96,0.13333333,0.55,0.01},
{0.8,0.02666667,0.575,0.01},
{0.7,0.34666667,0.5,0.01},
{0.64,0.33333333,0.7,0.01},{0.72,0.22222222,0.35,0.99},
{0.74,0.22222222,0.475,0.99},
{0,0.61111111,0.15,0.99},
{0.6,0.16666667,0.5,0.99},
{0.36,0.94444444,0.075,0.99},
{0.36,0.72222222,0.35,0.99},{0.56,0.47777778,0.275,0.99},
{0.4,0.90666667,0.3,0.99},
{0.6,0.44444444,0.225,0.99}};

Esta es la tabla cargada que se tiene sobre todos los patrones de entranemiento

* double pesosC1 [][] = newdouble [3][2];
Matriz de pesos entre capa de entrada y oculta
* double pesosC2 [] = new double [2];
Vector de pesos entre capa oculta de salida
* double salidaC2 [] = new double [2];
Vectorde Salida de la capa oculta
* double entradaC2 [] = new double [2]
Vector de entrada de la capa oculta
* double entradaCS,salidaK, dk, tasa=0.9;
entrada y salida de la capa de salida, asícomo la tasa y otros parámetros.
* double vectorE[]=new double [3];
Vector de pruebas

2. Métdo init:
Al ser un Applet, en esta clase se definen todos los componentes de la interfaz...
tracking img