Unit 3 - Pooling Layers

What is the primary purpose of pooling layers in a convolutional neural network (CNN)?

• To increase the number of parameters
• To reduce the spatial dimensions of feature maps
• To introduce non-linearity
• To add regularization to the network

Which of the following is a common type of pooling layer in CNNs?

• Dense Pooling
• ReLU Pooling
• Max Pooling
• Sigmoid Pooling

What does the "stride" hyperparameter in a pooling layer determine?

• The size of the pooling window
• The number of feature maps
• How much the pooling window moves
• The activation function used in pooling

Which pooling type is responsible for selecting the maximum value from a region of the input?

• Average Pooling
• Global Pooling
• Min Pooling
• Max Pooling

What happens to the spatial dimensions of the feature map after applying a pooling layer with a 2x2 pooling window and a stride of 2?

• They remain the same
• They double in size
• They halve in size
• They become four times larger

Global Average Pooling is often used in which part of a neural network architecture?

• Input layer
• Hidden layers
• Output layer
• Regularization layer

Which of the following is a disadvantage of using too much pooling in a CNN?

• Increased model complexity
• Slower training
• Loss of spatial information
• Improved generalization

Which of the following is an advantage of using Global Average Pooling (GAP) in a CNN?

• It reduces the number of parameters in the model.
• It retains fine-grained spatial information.
• It increases the computational complexity.
• It introduces non-linearity to the network.

In Global Max Pooling, what value is selected for a feature map?

• The average value of the entire feature map.
• The minimum value of the entire feature map.
• The maximum value from the entire feature map.
• The sum of all values in the feature map.

Which of the following is NOT a typical pooling window size in CNNs?

• 3x3
• 4x4
• 2x2
• 1x1