Pooling, also known as spatial pooling, is a method used to extract features in convolutional neural networks. It aggregates and statistically processes different features to obtain a relatively lower dimension while avoiding overfitting.

Pooling reduces the dimensions of each feature map while retaining most of the important information. Currently, there are mainly maximization, averaging, and summation.

Common pooling operations

The most common pooling operations are mean pooling and max pooling.

• Average Pooling:Calculate the average value of the image area and use it as the pooled value of the area.
• Max Pooling:Select the maximum value of the image area and use it as the pooled value of the area.

Define a spatial neighborhood and take the largest element from the modified feature map, or take the average.

The role of pooling

The pooling function can gradually reduce the spatial scale of the input representation:

• Reduce feature dimensions, and the number of parameters and calculations in the network can be reduced in a more controllable way;
• Make the network invariant to smaller changes, redundancies, and transformations in the input image;
• Assists in obtaining the maximum scale invariance of the image.