0
$\begingroup$

What would be considered a large (or medium or small) number of input neurons for a feedforward neural network?

While I am trying to do phoneme detection using inputs of 3200 sound-samples, I became curious how to categorize input sizes of a simple 3 layered ANN (input - hidden - output).

$\endgroup$
  • 1
    $\begingroup$ Why does it matter? You don't really have control over the size of your input layer. It will always have the same size as the number of your features. $\endgroup$ – zachdj Nov 19 at 14:07
  • $\begingroup$ I was just curious what the biggest feed forward nets were that have been successfully trained. And yes, I do have control in that I can decide the size of the inputs to a certain degree. $\endgroup$ – Skusku Nov 20 at 17:15
0
$\begingroup$

Can you provide more info about the context or what you are trying to do? I would say this depends heavily on the type of data you are inputting. Dealing with a simple 1280 x 720 pixels picture format will probably lead you to have at least 921600 input neurons.

$\endgroup$
  • $\begingroup$ I updated my post accordingly. It should be less broad now. $\endgroup$ – Skusku Nov 19 at 11:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.