AI Blog

AI Blog

by Michele Laurelli

Back to Glossary

Pooling Layer

/ˈpuːlɪŋ ˈleɪər/
Architecture
Definition

A downsampling layer in CNNs that reduces spatial dimensions while retaining important features.

Pooling layers reduce feature map size by aggregating neighboring values. Max pooling takes the maximum value, average pooling takes the mean. This provides translation invariance and reduces computation.

Examples

1

Max pooling 2x2

2

Average pooling

3

Global average pooling

Related Terms

CNN (Convolutional Neural Network)

A deep learning architecture specialized for processing grid-like data such as images, using convolutional layers.

Max Pooling

A pooling operation that takes the maximum value from each window of the feature map.

Michele Laurelli - AI Research & Engineering