Regarding the convolutional neural network pooling layer, the following description is correct?

Question

Options

• APooling operation is realized by scanning window
• BThe pooling layer can play a role in dimensionality reduction
• CCommonly used pooling methods include maximum pooling and average pooling
• DThe pooled feature image becomes smaller

• A
  100% (50)

Explanation

All four options accurately describe pooling layers in convolutional neural networks. A is correct because pooling works by sliding a scanning window (e.g., 2×2) across the feature map and applying an aggregation function within each window position. B is correct because pooling reduces the spatial dimensions (height and width) of feature maps, which lowers the number of parameters downstream - a form of dimensionality reduction. C is correct because max pooling (taking the maximum value in each window) and average pooling (taking the mean) are indeed the two dominant pooling strategies used in practice. D is correct because the output feature map after pooling is spatially smaller than the input - a 4×4 feature map with a 2×2 pool becomes 2×2.

There are no distractors here - this is an "all of the above" question where every statement is true.

Memory tip: Think of pooling as a "summarize and shrink" operation - it scans (A) a window, shrinks the image (D), reduces dimensions (B), and does so via max or average (C). All four facts describe the same mechanism from different angles.

No community discussion yet for this question.