How to deploy NARX or Neural Network in application builder

2 Ansichten (letzte 30 Tage)
Tina Zhao
Tina Zhao am 9 Jan. 2017
Kommentiert: Jeong_evolution am 10 Jun. 2017
I build a NARX to predict a time series prediction, and I was going to build an executable application so people who don't have Matlab can run it too. However, Matlab doesn't allow the direct inclusion of NARX or NET in the deployment. I followed the example, using getFunction to generate a function first. load dataset narx_net = narxnet(1:8,1:5,10); narx_net.divideFcn = ''; narx_net.trainParam.min_grad = 1e-10; [Xs,Xi,Ai,Ts] = preparets(narx_net,X,{},T); narx_net = train(narx_net,Xs,Ts,Xi); [yp1,xf,af] = narx_net(Xs,Xi,Ai); genFunction(narx_net,'narxnetFcn'); [y2,xf,af] = narxnetFcn(Xs,Xi,Ai);
However, in my application, I have to use this built NARX to my other test data, so this is how I use it: narx_net = narxnet(1:8,1:5,10); [p1,Pi1,Ai1,t1] = preparets(narx_net,u1,{},y1); [yp1,xf,af] = narxnetFcn(p1,Pi1,Ai1);
It seems I still need to use "narxnet" and "preparets" (functions from NN apps) to prepare the input data to be able to be used in narxnetFcn, so the application still has problem with deployment.
Anyone with past experience to successfully deployed them are welcome to advise.

Antworten (0)

Kategorien

Mehr zu Sequence and Numeric Feature Data Workflows finden Sie in Help Center und File Exchange

Community Treasure Hunt

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

Start Hunting!

Translated by