Is there any way to estimate how big the neural network would be after training session of 100,000 unlabeled images for unsupervised learning (like in STL-10 dataset: 96x96 pixels and color)?

Not the storage space (because this could vary I guess based on the implementation), but specifically how many neurons it could have. It could be an estimate (e.g. in thousand, millions). If it depends, then on what? Are there any figures that can be estimated?


1 Answer 1


The neural network is typically a set size once it's created. You'd have to create a network big enough for your data-set.

  • $\begingroup$ I don't see how this answers the question 'What is an estimate of how many neurons it would have?' $\endgroup$
    – Mithical
    Aug 25, 2016 at 18:46
  • 5
    $\begingroup$ I think Zakk's point was that training does not affect the size of a neural network. You define the size when you create it. How many times you train it, or what the size of the data set is, doesn't change the size of the network once it's defined. Well, at least not using any technique I'm familiar with. $\endgroup$
    – mindcrime
    Aug 25, 2016 at 19:00
  • 5
    $\begingroup$ This certainly answers the question of how many neurons the net would have. It would have the same number of neurons that it started with, and that number is whatever number the operator picked. $\endgroup$ Aug 30, 2016 at 3:29

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .