0
$\begingroup$

I've done a CNN project with Keras and OpenCV, and I've got roughly 65% accuracy. And now I have to present this work in my University, but I'm afraid if the teachers ask me for how do I knew how to construct the right layers to my CNN.

In fact in my development I took a look in others projects to see how the other people do, and I know that you have to have a number of neurons equivalent to the number and size of your image. But I know that has a mathematic behind this.

So how could I tell if they ask me it?

$\endgroup$
0
$\begingroup$

There isn't a formula behind it - layer sizes are selected with a lot of trial and error (often automated by cross-validation). But saying "I've used an architecture from a network from a similar problem, or at least from the same domain" is a legitimate answer.

$\endgroup$

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.