Problem in training neural network to make a standalone application in Matlab 2016?

1 次查看(过去 30 天)
Hi, I use the 'mcc -m funcname.m' to make a standalone application. There is no problem in simple case. I use this command and standalone applications work well. BTW, in using neural network generate error. I have a function include a neural network; This function loads data, trains and predict. standalone application is created successfully, but when I run the application as the standalone application, it generates error message. The error is:
"Default value is not a member of type "nntype.training_fcn". Error using nnetParamInfo (line 28) FCN is not the name of a function on the MATLAB path. Error in layrecnet>get_info (line 96) Error in layrecnet (line 46) Error in UsingRecurrent (line 23) nnet:nnetParamInfo:Arguments"
Is it possible to compile neural network train in Standalone application (Matlab 2016)? If yes, how can I don?
Thank you in advanced.

回答(2 个)

Marc Jakobi
Marc Jakobi 2016-10-7
I don't think you can train a neural network in a standalone. It should be possible to train the network in Matlab, save it in a mat file and create a standalone that loads the mat file and uses the network on new data though.
  5 个评论
Ali Aryan
Ali Aryan 2016-10-13
Thank you Mr. Marc Jakobi for your interest and help. I think my question was not clear enough. Actually I am using Matlab program as an engine program for a web site. so, i decided to use Standalone application. Then, because of above explained problem, I changed the host to a Matlab support host.
Marc Jakobi
Marc Jakobi 2016-10-13
Ah, now I get it. I thought you were calling the bat file from within the standalone just to train the network.

请先登录,再进行评论。


Walter Roberson
Walter Roberson 2017-6-10
The ability to deploy applications that train a neural network was added in R2016b.

类别

Help CenterFile Exchange 中查找有关 Deep Learning Toolbox 的更多信息

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by