I am currently using Matlab 2012rb, and when i use the following command "NEWFF" for feed-forward back propagation network in command window i get following warning :
Warning: NEWFF used in an obsolete way.
> In obs_use at 18
In newff>create_network at 127
In newff at 102
and my program does not execute...please tell me what is an alternative command, i googled alot but did not find any alternative.
Use the appropriate syntax for NEWFF (or its replacement) for your version
of MATLAB. In your MATLAB session, type:
doc newff
and you should see the version of that example appropriate for the NEWFF
included with your version's installation of Neural Network Toolbox. Do
this unless you're using the latest release, release R2010b, in which case
you should use FEEDFORWARDNET as described in the Release Notes for this
version:
http://www.mathworks.com/help/toolbox/nnet/rn/rn2.html#92388
If this is not an option, you _could_ use WARNING to suppress the warning,
but that's a band-aid fix. I strongly encourage you to use the other
approach.
it seems that the "head" of the function has changed (that'st why matlab shows the "obsolete" message). May be this link can help you:
http://stackoverflow.com/questions/9559108/how-to-set-newff-neural-network-toolbox-function-in-matlab
[]s
Hi, try to use "fitnet" instead of "newff". It does the same function. Sure, it will work. Good luck.
Y use the command when you can use the GUI...go to start and from the toolbox, select the NNtool...
Use the appropriate syntax for NEWFF (or its replacement) for your version
of MATLAB. In your MATLAB session, type:
doc newff
and you should see the version of that example appropriate for the NEWFF
included with your version's installation of Neural Network Toolbox. Do
this unless you're using the latest release, release R2010b, in which case
you should use FEEDFORWARDNET as described in the Release Notes for this
version:
http://www.mathworks.com/help/toolbox/nnet/rn/rn2.html#92388
If this is not an option, you _could_ use WARNING to suppress the warning,
but that's a band-aid fix. I strongly encourage you to use the other
approach.
You have to use "feedforwardnet" for that but its better to use 'fitnet' and 'patternet' as they both have built in feedforwardnet. The first issue for you will be to arrange your data in the form of row matrix or column matrix. If you do not have the data in proper form, you can never run the network. The formation of data depends on the application and type of problem that you are facing. run the fitnet example and then check it for your data. I hope that works for u
- Badges
- Science topic