Compare convolutional neural networks (CNNs) with fully connected (dense) networks.
Explain:
-
The structural differences between convolutional layers and fully connected layers (parameter sharing, locality, connectivity pattern).
-
Why CNNs are more efficient for certain data types.
-
What kinds of
data or tasks
are most suitable for CNNs, and when a fully connected architecture is more appropriate.