Check your understanding of this module before moving on.
?
0/3
Checkpoint quiz 3 questions
A quick check on this module's key ideas before you move on
Question 01
What is the general design pattern shared by classic CNN architectures like LeNet, VGG, and ResNet for image classification?
A is correct. Most classic CNNs follow a funnel shape: repeated conv+activation (often with pooling) blocks that shrink spatial size while growing the number of channels/features, ending in fully connected layers that produce final class scores.
Question 02
What problem did residual (skip) connections in ResNet specifically address?
A is correct. Very deep plain networks were found to perform worse than shallower ones, partly due to optimization difficulty. Skip connections give gradients (and the original signal) a direct path around a block, making very deep networks much easier to train successfully.
Question 03
Why do CNN architectures typically increase the number of channels (feature maps) while decreasing spatial resolution as you go deeper?
A is correct. As you move deeper, the network trades fine spatial detail (which matters more for low-level edges and textures) for a richer set of abstract feature channels (which matters more for recognizing complex objects), matching how visual information becomes more abstract at higher levels.