How deeply does Deep Learning with R, Third Edition cover convolutional layer?

Explains patch-based feature maps, output channels, padding and strides through architecture and shape examples, without a formal derivation.

Convolutional layer

As explained in Deep Learning with R, Third Edition

A convolutional layer is a neural-network layer that applies learned local filters across spatial or sequential input dimensions.

A convolutional layer learns local patterns in an image instead of looking at every pixel at once. It takes small patches from a feature map and applies the same learned transformation at each location. A pattern learned in one part of an image can therefore be recognized elsewhere. Stacking these layers lets a network build from simple patterns, such as edges, toward more complex visual features. section 8.1.1

Listing 8.1 Instantiating a small convnet

library(keras3)

inputs <- keras_input(shape = c(28, 28, 1))
outputs <- inputs |>
  layer_conv_2d(filters = 64, kernel_size = 3, activation = "relu") |>
  layer_max_pooling_2d(pool_size = 2) |>
  layer_conv_2d(filters = 128, kernel_size = 3, activation = "relu") |>
  layer_max_pooling_2d(pool_size = 2) |>
  layer_conv_2d(filters = 256, kernel_size = 3, activation = "relu") |>
  layer_global_average_pooling_2d() |>
  layer_dense(10, activation = "softmax")

model <- keras_model(inputs = inputs, outputs = outputs)

This R/Keras model starts with 28 × 28 grayscale digit images. Three convolutional layers alternate with pooling layers; global average pooling and a softmax layer turn the resulting features into ten digit probabilities. section 8.1

Read each `layer_conv_2d()` call in terms of its input and output:

The input is a feature map with height, width and channels. An MNIST image has one channel; an RGB image has three. section 8.1 section 8.1.1
`kernel_size = 3` selects 3 × 3 patches. The `filters` argument sets the number of output channels, each representing a learned pattern’s response across the image. section 8.1 section 8.1.1
Height and width can shrink as layers accumulate. Padding and strides affect which patches are used and how the spatial dimensions change; pooling is another way to downsample. section 8.1.1 section 8.1

Classify handwritten digits

Reshape MNIST images to `(28, 28, 1)`, divide pixel values by 255, then compile the model with `rmsprop`, `sparse_categorical_crossentropy` and accuracy. Fit for five epochs with batches of 64 and evaluate on the test images. The reported test accuracy is 0.9931, compared with 97.8% for the earlier densely connected model. section 8.1

Choose downsampling to suit the output. For classification, pooling can reduce feature-map size before a final classifier. For image segmentation, pixel locations matter: strided convolutions downsample while retaining more location information than max pooling, and transposed convolutions restore spatial dimensions for per-pixel predictions. section 8.1 section 11.3.2

How Deep Learning with R, Third Edition covers it

Cover of Deep Learning with R, Third Edition

liveBook manning.com

Read Deep Learning with R, Third Edition on liveBook Get the book at manning.com

Also explained in

liveBook manning.com

Natural Language Processing in Action, Second EditionA convolutional layer is a neural-network layer that applies learned filters across local regions of its input to produce feature maps.
liveBook manning.com

Math and Architectures of Deep LearningA neural-network layer that applies convolution operations to an input, commonly used to extract local features from images.
liveBook manning.com

Deep Learning with PyTorch, Second EditionA PyTorch 2D convolution layer module that applies learned filters over spatial input regions.
liveBook manning.com

Deep Learning with Python, Third EditionA two-dimensional convolution layer used here with a 1×1 kernel to project a residual to a matching shape.
liveBook manning.com

Grokking AI Algorithms, Second EditionA convolutional layer is a neural-network layer that applies shared filters to local regions of an input representation while preserving spatial organization.
liveBook manning.com

Evolutionary Deep Learning: Genetic algorithms and neural networksA convolutional layer applies learned local filters across structured input locations to detect spatial or otherwise local patterns.
liveBook manning.com

Optimization Algorithms: AI techniques for design, planning, and control problemsA transformation that applies shared learned weights to neighboring feature representations.
liveBook manning.com

Learn Generative AI with PyTorchA PyTorch convolution layer that applies learned 2D filters to an input tensor to produce feature maps.
liveBook manning.com

Sutskever's ListA convolutional layer is a neural-network layer that applies learned local filters across an input to produce spatial feature responses.
liveBook manning.com

Machine Learning Algorithms in DepthA convolutional layer is a neural-network layer that applies learned convolution kernels to an input representation.
liveBook manning.com

Build a Text-to-Image Generator (from Scratch)Conv2d is a two-dimensional convolution layer that applies learned spatial filters to image-like tensors.
liveBook manning.com

Distributed Machine Learning PatternsA convolutional layer is a neural-network layer that applies learned convolutional filters to its input.
liveBook manning.com

Grokking Machine LearningA convolutional layer is a neural-network layer that produces a filtered representation by applying convolutions across an input.

Compare how 14 books explain convolutional layer →

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.
feature mapproduces · A feature map is a spatially organized array of neural-network activations representing the presence or strength of learned features at different locations.
downsamplingused by · Downsampling is the reduction of data quantity, spatial resolution, temporal resolution, or sampling rate by retaining, selecting, or aggregating fewer representative values or observations.
convolution kerneluses · A convolution kernel is a small array or tensor of weights applied across local input regions to compute weighted responses and detect or transform patterns.
batch sizeuses · Batch size is the number of items grouped together for processing or computation as a single batch.
encoderused by · An encoder is a component that transforms input data into an encoded representation for subsequent processing, reconstruction, comparison, or transmission.

This explainer was generated by Manning's knowledge graph from the text of Deep Learning with R, Third Edition. The section links point to the parts of the book it draws on.