newff function does not appear in 2023b version

Hello! I'm starting to learn MatLab in university and I have a problem. My teacher gave us a formula that starts with:

nn1 = newff(minmax(input),[3,1],{'tansig','purelin'},'trainlm');

Everytime that I run the code it says that newff doesn't function in this version anymore.

Can anyone help me to rewrite that line of code please, I really don't know anything about the program.

I have the 2023b student version.

Answer 1

[Walter Roberson]

No, the last time newff was documented was R2010a. The teacher needs to rewrite the lab.

See fitnet and feedforwardnet

https://www.mathworks.com/matlabcentral/answers/35772-what-is-the-difference-between-feedforwardnet-with-fitnet

[Steven Lord]

It still exists in Deep Learning Toolbox, but Walter is correct that it was last documented/used in release R2010a (back when that toolbox was named Neural Network Toolbox.)

which -all newff

/MATLAB/toolbox/nnet/nnet/nnnetwork/newff.m

dbtype 1:5 newff.m

1     function out1 = newff(varargin)
2     %NEWFF Create a feed-forward backpropagation network.
3     %
4     %  Obsoleted in R2010b NNET 7.0.  Last used in R2010a NNET 6.0.4.
5     %  The recommended function is feedforwardnet.

If you receive an "undefined function" error, you likely don't have Deep Learning Toolbox installed. To check this, use the ver function. If it doesn't list Deep Learning Toolbox, neither newff nor the replacements Walter suggested will be available.

ver nnet

-----------------------------------------------------------------------------------------------------
MATLAB Version: 23.2.0.2436196 (R2023b) Update 4
MATLAB License Number: 0
Operating System: Linux 5.4.233-0504233-generic #202302250651 SMP Sat Feb 25 12:26:27 UTC 2023 x86_64
Java Version: Java 1.8.0_292-b10 with AdoptOpenJDK OpenJDK 64-Bit Server VM mixed mode
-----------------------------------------------------------------------------------------------------
Deep Learning Toolbox                                 Version 23.2        (R2023b)

which -all fitnet

/MATLAB/toolbox/nnet/nnet/nnnetwork/fitnet.m

which -all feedforwardnet

/MATLAB/toolbox/nnet/nnet/nnnetwork/feedforwardnet.m

But I second Walter's recommendation to recommend the professor update the lab exercise. There have been a lot of improvements to Deep Learning Toolbox in the 13+ years since that lab was written.

[María]

Thank you!!