多层神经网络实现

 

matlab实现多层神经网络训练

激活函数：

function [y] = sigmod(x)
    % y = 1.0 ./ (1 + exp(-x)); % sigmod函数
    y = 2.0 ./ (1 + exp(-2.*x)) - 1; % tansig函数
end

 激活函数导数：

function [y] = sigmod_deri(x)
    % y = x .* (1 - x);
    y = 1 - x.^2;
end

多层神经网络训练： 

clear;
clc;
close all;

% 构造数据
x = 1 : 0.01 : 10;
y = 1.1 * x.^3 + 2.1 * x.^2 + 3.1 * x + 1.2 + 15 .* sin(100.*x);
len = length(x);

% 数据归一化
map_min = -1;
map_max = 1;
min_x = min(x);
max_x = max(x);
min_y = min(y);
max_y = max(y);
normal_x = (x - repmat(min_x, 1, len)) ./ (max_x - min_x) .* (map_max - map_min) + map_min;
normal