annealing.cpp

BP神经网络算法的VC＋源码

/*
    nn-utility (Provides neural networking utilities for c++ programmers)
    Copyright (C) 2003 Panayiotis Thomakos

    This library is free software; you can redistribute it and/or
    modify it under the terms of the GNU Lesser General Public
    License as published by the Free Software Foundation; either
    version 2.1 of the License, or (at your option) any later version.

    This library is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
    Lesser General Public License for more details.
*/
//To contact the author send an email to [e-mail]


#include <nn-utility.h>

using namespace nn_utility;
//define a floating-point handler for "nn_utility_functions"
nn_utility_functions<float> derived;

int main(){
	//create a widrow hoff network layer named "hidden1"
	WIDROW_HOFF<float> *hidden1 = new WIDROW_HOFF<float>( );
	//define it's size 2x2 and it's elements
	hidden1->definef( 2, 2, 2.0, 2.0, -1.0, -1.0 );

	//create an initial input vector called "IN"
	nn_utility_functions<float>::VECTOR IN;
	derived.LoadVectorf( IN, 2, 1.0, 1.0 );
	
	//define "hidden1" as an annealing network (see MANUAL for details)
	hidden1->set_annealing( IN, 2, 1.0, 0.8, 5 );

	//create a buffer for "hidden1", name the buffer "ppHidden1"
	layer<float> *ppHidden1 = hidden1;

	//train the network for 100 interations as 0.1 learning rate, and present output
	derived.train( &ppHidden1, 100, 0.1, true );

	//Present output:
	cout << "Simulated Annealing found minima to be ";
	derived.PrintVector( ppHidden1->annealing_input, 2 );

	cout << "Notice in the output that the network has clearly converged on the\n";
	cout << "minima of the network function. The minima should be [ 0 1 ].\n";
	//attempts to find value of f(x) : f(x) < than all others by localizing minima
	cout << '\n';
	return 0;
}