How do i use sequentialfs in matlab when using neural networks (patternnet) as a model?

Question

Rayan Fayad 2019-7-10

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/471010-how-do-i-use-sequentialfs-in-matlab-when-using-neural-networks-patternnet-as-a-model

回答： Divya Gaddipati 2019-7-18

%after loading the data
c = cvpartition(y,'k',10);
opts = statset('display','iter');
classf = @(xtrain,ytrain,xtest,ytest)sum(classify(ytest,xtest,train(ytrain,xtrain)) ~= ytest); 
[fs,history] = sequentialfs(classf,PD_Inputs,y,'cv',c,'options',opts)

Its not working and im getting this error:

Error using crossval>evalFun (line 480)
The function
'@(xtrain,ytrain,xtest,ytest)sum(classify(ytest,xtest,train(ytrain,xtrain))~=ytest)' generated
the following error:
Undefined function 'train' for input arguments of type 'double'.
Error in crossval>getFuncVal (line 497)
funResult = evalFun(funorStr,arg(:));

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

Divya Gaddipati 2019-7-18

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/471010-how-do-i-use-sequentialfs-in-matlab-when-using-neural-networks-patternnet-as-a-model#answer_383810

Hi,

classf = @(xtrain,ytrain,xtest,ytest) sum(classify(ytest,xtest,train(ytrain,xtrain)) ~= ytest); 

In the above declaration, you used the classify function from the Statistics and Machine Learning Toolbox which is https://www.mathworks.com/help/stats/classify.html

Since, you want to use a neural network, you should be using the classify and train functions from the Deep Learning Toolbox. The train function outputs the trained shallow neural network. The classify function returns the predicted labels and scores (if required).

For more information on how to use these functions, you can refer to the following links:

https://www.mathworks.com/help/deeplearning/ref/classify.html

https://www.mathworks.com/help/deeplearning/ref/train.html

Also, since you want to use patternnet, one possible modification which can be made to your code is:

classf = @(xtrain, ytrain, xtest, ytest) 
        sum(classify(train(patternnet(50), xtrain, ytrain), xtest) ~= ytest); 

You can also use the trainNetwork function which allows you to model your own deep neural networks or load any pre-trained network. For more information, you can refer to the following link:

https://www.mathworks.com/help/deeplearning/ref/trainnetwork.html