AI Blog

AI Blog

by Michele Laurelli

Back to Glossary

Max Pooling

/mæks ˈpuːlɪŋ/
Technique
Definition

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

Max pooling reduces spatial dimensions while preserving the strongest activations. Provides translation invariance and reduces computation. Most common pooling method in CNNs.

Examples

1

2x2 max pooling

2

Downsampling by factor 2

3

Spatial invariance

Related Terms

Pooling Layer

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

CNN (Convolutional Neural Network)

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

Michele Laurelli - AI Research & Engineering