In MATLAB , NANs in during training neural network

I encountered a very strange problem while using Matlab 2016a Using a neural network built into the Perceptron NNtool toolbox, I try to predict a nonlinear variable. However, during training, the values obtained are for testing and training and validation of NANs. I tried to change the learning speed and hide the layers, but nothing worked. Do you have any idea, why is this so? Any help is greatly appreciated!

[Arjun Suhass]

Please improve this question with a reproducible code sample or link to the example.

Answer 1

[David Willingham]

Hi,

MathWorks didn't build the Perceptron NNtool toolbox. You can run shallow neural networks in both the Statistics and Machine Learning toolbox:

https://www.mathworks.com/help/stats/neural-networks-for-classification.html

and within Deep Learning Toolbox there is also functionality:

https://www.mathworks.com/help/deeplearning/gs/classify-patterns-with-a-neural-network.html

Have you had a look at either of those? If so, then there is support MathWorks could offer