How does using Convolutions in our Deep neural network impact training?
1)It does not affect training
2)It makes it faster
3)Its impact will depend on other factors.
4)It makes it slower
Answer->
In some cases, a plain DNN will be faster, so It makes it faster
But shouldn’t the answer be 'It makes it slower?
Please explain…
Convolution layer reduces the number of parameters required since it summarizes pixels within its neighborhood.
If you’ve taken up deep learning specialization, please watch this lecture keeping impact of convolution on the number of parameters in mind.
but does it will make it slower in the training? but faster in predict? cmiiw
I don’t understand your question.