Is there some way to display the output FUNCTION of a neural net (i.e. the actual equation relating the inputs to the neural net output)?

I have generated some neural networks using the neural network fitting app in Matlab, and would like to see an equation relating the inputs of the network to the resulting model (like a y=mx+b kind of output). I see that whenever I generate the regressions for the model, it shows its relationship to the targets, but is there some way to see the model's relationship with the inputs?

Answer 1

[Greg Heath]

I have several posts regarding this in both the NEWSGROUP and ANSWERS.

It should suffice to search in ANSWERS using

 GREG ANALYTIC

Hope this helps.

Thank you for formally accepting my answer

Greg

Answer 2

[Steven Lord]

Theoretically, yes, you could retrieve the weights, biases, functions, etc. used in the network from its properties and write the equations. But it's not going to be quite as simple as "y = m*x+b" even in the simple one neuron case; see this example in the documentation. You're more likely to have something like the multiple neuron example on this documentation page instead. And those don't take into account the input- and output-processing functions.

[Zohaib Moonis]

I got that haha, that was just the simplest way of communicating my thoughts. Thank you!