1
$\begingroup$

When training a cnn network, only the filters are updated, but how are the Fully Connected Layer weights updated?

$\endgroup$
1
$\begingroup$

Their weights are also trained as well as the weights of the CNN filters.

So, the gradient of the loss is computed with respect to these weights also, and at each gradient step, both of the dense (fully connected) and CNN layers' weights are updated.

$\endgroup$

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.