How to use weights and bias from Neural Network Toolbox for classification

Question

David Franco 2020-8-30

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/586310-how-to-use-weights-and-bias-from-neural-network-toolbox-for-classification

回答： David Franco 2020-10-1

I found a code in Matlab Answers for regression:

x = 0:0.1:10; % input dimension 1x101
y = sin(x); % output dimension 1x101
neurons = 10;
net = feedforwardnet(neurons);
net = train(net,x,y);
outputs = sim(net,x);
w1 = net.IW{1,1};
w2 = net.LW{2,1};
b1 = net.b{1};
b2 = net.b{2};
xx = x;
yy = zeros(size(y));
for ii = 1:length(net.inputs{1}.processFcns)
    xx = feval(net.inputs{1}.processFcns{ii},...
        'apply',xx,net.inputs{1}.processSettings{ii});
end
for jj = 1:size(xx,2)
    yy(jj) = w2*tansig(w1*xx(jj) + b1) + b2;
end
for kk = 1:length(net.outputs{2}.processFcns)
    yy = feval(net.outputs{2}.processFcns{kk},...
        'reverse',yy,net.outputs{2}.processSettings{kk});
end

How can I change the following equation for a classification problem, for example [x,y] = simpleclass_dataset?

yy(jj) = w2*tansig(w1*xx(jj) + b1) + b2;

Because the input and output dimensions are different from the regression example.

For simpleclass_dataset the input dimension is 2x1,000 and the output dimension is 4x1,000.

Thank you!

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

David Franco 2020-10-1

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/586310-how-to-use-weights-and-bias-from-neural-network-toolbox-for-classification#answer_503314

在 MATLAB Online 中打开

I found the answer (output and yy are equal):

% Simulated a Neural Network
clear
close
clc
rng default
[x,y] = simpleclass_dataset;
neurons = 10;
net = feedforwardnet(neurons);
net = train(net,x,y);
outputs = sim(net,x);
% outputs = round(outputs);
figure, plotconfusion(y,outputs)
w1 = net.IW{1,1};
w2 = net.LW{2,1};
b1 = net.b{1};
b2 = net.b{2};
xx = x;
for ii = 1:length(net.inputs{1}.processFcns)
    xx = feval(net.inputs{1}.processFcns{ii},...
        'apply',xx,net.inputs{1}.processSettings{ii});
end
a1 = tanh(w1*xx + b1);
yy = purelin(w2*a1 + b2);
for mm = 1:length(net.outputs{1,2}.processFcns)
    yy = feval(net.outputs{1,2}.processFcns{mm},...
        'reverse',yy,net.outputs{1,2}.processSettings{mm});
end
figure, plotconfusion(y,yy)