1
$\begingroup$

I was wondering if there are any rules regarding correlation between the predictor variables or between the predictor and outcomes variables of a CNN. What should be the value of correlation? Does it matter when working with CNN? I have the case of using two categorical predictor variables and

$\endgroup$
1
  • $\begingroup$ You may want to search for neural style transfer networks. $\endgroup$ Jun 25, 2019 at 9:21

1 Answer 1

1
$\begingroup$

In theory, convolutional and pooling layers should be able to account for this kind of collinearity. Convolutional filters will learn to extract the relevant information from your pixel data, and pooling layers significantly reduce their dimensionality, reducing multicollinearity as a side effect.

However, if you think multicollinearity is still a problem, you can apply some dimensionality reduction techniques. They work on image data as well!

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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