Pooling

As explained in Build a Text-to-Image Generator (from Scratch)

A pooling layer aggregates values within local spatial windows to produce a reduced or transformed feature representation.

Pooling aggregates values inside local spatial windows to reduce feature-map dimensions and produce more compact representations. In convolutional networks, it improves computational efficiency while preserving prominent features. section 6.5 section 6.2.3

A max-pooling operation proceeds in four stages:

• Divide the image or feature map into local windows, such as 2 × 2 or 3 × 3 patches. section 6.2.3
• Set the stride to determine how far the window moves between operations; a stride of 2 advances two pixels horizontally and vertically. section 6.2.3
• Select the largest value in each window and use it as that window’s output. section 6.2.3
• Place the reduced feature map in the contracting path, where repeated convolution and pooling blocks capture increasingly abstract features. section 6.2.2

Max pooling in PyTorch

For a 4 × 4 single-channel tensor, torch.nn.MaxPool2d(2) applies a 2 × 2 window with stride 2 and produces [[2, 4], [5, 7]]. section 6.2.3

Figure 6.5 A numerical example of how max pooling works, with a kernel size of 2 × 2 and a stride equal to 2

Inspect the numerical window-by-window calculation behind 2 × 2 max pooling with stride 2. section 6.2.3

How Build a Text-to-Image Generator (from Scratch) covers it

How deep it goes

Explains windows, kernels, strides, maxima, and a PyTorch calculation, but does not develop pooling gradients or padding.

Examples

Uses a 4 × 4 tensor, 2 × 2 windows, stride 2, and the output [[2, 4], [5, 7]].

What it assumes

Assumes basic images, feature maps, convolutional layers, and tensor shapes while teaching pooling windows and strides explicitly.

What you can do after

Apply it via PyTorch’s MaxPool2d layer to downsample image tensors and verify the resulting values.

How it connects

Connects pooling with convolutional layers, U-Net encoder-decoder paths, bottlenecks, transposed convolutions, skip connections, and average pooling.

liveBookmanning.com

Read Build a Text-to-Image Generator (from Scratch) on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Inside Deep Learning: Math, Algorithms, ModelsA neural-network operation that downsamples local neighborhoods by aggregating their values.
• 

  liveBookmanning.com
  Natural Language Processing in Action, Second EditionA pooling layer is a neural-network layer that aggregates neighboring activations to reduce their representation while retaining salient information.
• 

  liveBookmanning.com
  Math and Architectures of Deep LearningA downsampling operation on feature maps that reduces spatial resolution, often by local averaging or max selection.
• 

  liveBookmanning.com
  Evolutionary Deep Learning: Genetic algorithms and neural networksPooling is a neural-network operation that aggregates values within local regions to reduce spatial or feature-map dimensions.
• 

  liveBookmanning.com
  Graph Neural Networks in ActionA graph operation that reduces a collection of representations to a smaller representation, often for graph-level prediction.
• 

  liveBookmanning.com
  Deep Learning with PyTorch, Second EditionA downsampling technique that reduces spatial resolution by aggregating local regions of a feature map.
• 

  liveBookmanning.com
  AI-Powered SearchPooling is the aggregation of multiple representation values into a single lower-level or fixed-size representation.

Compare how 8 books explain pooling →

Related concepts

• convolutional neural networkused by · A convolutional neural network is a neural-network architecture that applies learned convolutional operations to extract patterns from structured data.
• max poolingincludes · Max pooling is an aggregation operation that replaces each local or grouped region with its maximum value, often to downsample a representation.
• downsamplingequivalent to · Downsampling is the reduction of data quantity, spatial resolution, temporal resolution, or sampling rate by retaining, selecting, or aggregating fewer representative values or observations.
• upsamplingalternative to · Upsampling is a process that increases the number, density, or spatial or temporal resolution of samples or values in a dataset or representation by adding, duplicating, or estimating them.
• gradientuses · A gradient is the vector of partial derivatives of a scalar function with respect to its variables, indicating its local rate and direction of change.

This explainer was generated by Manning's knowledge graph from the text of Build a Text-to-Image Generator (from Scratch). The section links point to the parts of the book it draws on.