You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Sathiyanarayanan S edited this page Jan 12, 2021
·
4 revisions
Why Rectified Linear Units is a good activation function?
For Alexnet we usually pass 224224 images (227227 with padding), What happens if we pass image with less resolution (pixel) / more resolution? Does it fail / overfit / poor accuracy?
With fully connected layer it will fail even if the resolution changes by 1 pixel. Because at the point of flattening number of nodes expected by fully connected layer will not match.