在打包 MATLAB App 时，appdesig​ner设计中包含神经​网络相关程序，但是打​包出来却没有用，是为​什么

[f,p]=uigetfile({'*.xlsx;*.txt;*.xls;','All Data Files'});

if (ischar(p))

fname = [p f];

A=xlsread(fname);

app.UITable.Data=A;

end

% A=xlsread('test1.xlsx');

% app.UITable.Data=A;

input=A(1:end,1:5);

output=A(1:end,6:7);

insize=size(input,2);

outsize=size(output,2);

%% 划分训练集和测试集

num=size(input,1);

rowrankA = randperm(num); % 随机打乱的数字，从1~行数打乱

input= input(rowrankA, :);%按照rowrank打乱矩阵的行数

output=output(rowrankA, :);

Xtrain=input(1:floor(0.8*num),:);

Ytrain=output(1:floor(0.8*num),:);

Xtest=input(floor(0.8*num)+1:num,:);

Ytest=output(floor(0.8*num)+1:num,:);

%% 训练集归一化

muX = mean(Xtrain,1);

sigmaX = std(Xtrain,0,1);

muT = mean(Ytrain,1);

sigmaT = std(Ytrain,0,1);

Xtrain = (Xtrain - muX) ./ sigmaX;

Ytrain = (Ytrain - muT) ./ sigmaT;

%% 创建网络结构

layers = [

featureInputLayer(insize,"Name","featureinput")

fullyConnectedLayer(10,"Name","fc")

fullyConnectedLayer(outsize,"Name","fc_1")

regressionLayer("Name","regressionoutput")];

%% 设置训练选项

Xtest = (Xtest - muX) ./ sigmaX;

Ytest = (Ytest - muT) ./ sigmaT;

b={Xtest,Ytest};

options = trainingOptions("adam", ...

MaxEpochs=500, ...

Shuffle="every-epoch", ...

Plots="training-progress", ...

Verbose=0 ,...

ValidationData=b);

%% 训练网络

net = trainNetwork(Xtrain,Ytrain,layers,options);

% X1=app.EditField_2.Value;

% X2=app.EditField.Value;

% X3=app.EditField_3.Value;

% X4=app.EditField_4.Value;

% X5=app.EditField_5.Value;

X=[1,1,1,1,1];

X = (X - muX) ./ sigmaX;

Y= predict(net,X);

Y= Y.*sigmaT+muT;

Y=double(Y);

代码在脚本中是这样的

Answer 1

[Umang Pandey]

Hi Zhi,

You can refer to the following MATLAB answers for more information on how to package your Neural Network app -

https://www.mathworks.com/matlabcentral/answers/444474-neural-network-in-matlab-app-designer

Hope this helps!

Best,

Umang