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 does a max pooling layer do to its input feature map?
A is correct. Max pooling slides a small window (e.g. 2x2) across the feature map and outputs just the largest value in each window, which both downsamples the map and keeps the most salient activation from each local region.
Question 02
Why does pooling help make a CNN somewhat invariant to small translations of features in the input?
A is correct. If a strong edge shifts slightly within a pooling window, the max value in that window is likely to stay the same. This local insensitivity to small shifts is what gives the network some robustness to minor translations in the input.
Question 03
What is a feature map, in the context of a trained CNN's intermediate layers?
A is correct. Each filter in a convolutional layer produces its own 2D grid of outputs as it slides across the input. This grid, the feature map, shows where in the image that particular filter's pattern was strongly present.