Convolution kernel
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.
Natural Language Processing in Action, Second Edition
A convolutional kernel is a parameter matrix or tensor whose sliding weighted operation detects a local pattern in an input sequence.
- How it is used
- Detects local patterns in ordered text representations and supplies features for convolutional text classification.
- How deep it goes
- Works from a stencil analogy through weighted Python loops, multichannel PyTorch weights, and kernel-dependent output shapes.
- Examples
- Uses an adjacent-value average, an adverb–verb–noun match, and a handcrafted Morse-code dot detector.
- What you can do after
- Build it from scratch as a sliding weighted calculation, then apply it via PyTorch Conv1d to multichannel text features.
Read how Natural Language Processing in Action, Second Edition explains convolution kernel →
Math and Architectures of Deep Learning
A learned kernel applied in a convolution layer to detect local spatial patterns in images.
- How it is used
- Captures local patterns in arrays, images, and video by weighting neighboring values during convolution.
- How deep it goes
- Derives sliding weighted sums and output sizing, then relates convolution to matrix multiplication and backpropagation.
- Examples
- Uses moving-average curves, edge-detection signals, noisy text images, image boundaries, and a moving-ball video.
- What you can do after
- Apply it via PyTorch Conv1d, Conv2d, and Conv3d by setting kernel weights and tensor shapes for demonstrations.
Read how Math and Architectures of Deep Learning explains convolution kernel →
Inside Deep Learning: Math, Algorithms, Models
A learned array of weights applied over local input regions to compute convolution outputs.
- How it is used
- Applies local, shared weights to encode spatial relationships and detect useful patterns in image tensors.
- How deep it goes
- Defines shapes, sliding arithmetic, channels, padding, and multiple filters, but defers full convolution-layer theory.
- Examples
- Uses 1D difference detection, 2D numeral images, blur and edge kernels, MNIST, and PyTorch convolution layers.
- What you can do after
- Build it from scratch with nested loops, choose padding and kernel shapes, and configure multiple filters with PyTorch.
Read how Inside Deep Learning: Math, Algorithms, Models explains convolution kernel →
Deep Learning with PyTorch, Second Edition
A small learned filter window applied over local regions of an input to compute weighted sums.
- How it is used
- Uses shared kernel weights to detect local image patterns wherever they occur.
- How deep it goes
- Works through a 3 × 3 weighted sum, multichannel weight shapes, and learned versus hand-set filters; leaves continuous convolution aside.
- Examples
- Demonstrates RGB-image convolution, a neighborhood-averaging blur, and a vertical-edge detector.
- What you can do after
- Apply it via PyTorch `nn.Conv2d` to specify kernel size and channel counts, and inspect the resulting weight shape.
Read how Deep Learning with PyTorch, Second Edition explains convolution kernel →
Grokking AI Algorithms, Second Edition
A convolutional filter is a small learned grid of weights that detects a particular local pattern in image data.
- How it is used
- Extracts local image features in convolutional layers and supports learned upsampling in a U-Net decoder.
- How deep it goes
- Works through sliding dot products and feature-map shapes, then sketches NumPy implementations; does not derive weight updates.
- Examples
- Uses a numeric filter-row calculation and eight 3×3 filters scanning an 8×8 noisy image.
- What you can do after
- Build it from scratch using NumPy-style loops to scan image patches and calculate feature maps.
Read how Grokking AI Algorithms, Second Edition explains convolution kernel →
Deep Learning with Python, Third Edition
The learned weight tensor used by a convolution to transform local input patches.
- How it is used
- Explains how convolutional layers apply learned local transformations and produce spatial response maps for image classification.
- How deep it goes
- Defines patch size, filter count, and response maps, then works through filter visualization without deriving kernel-weight calculations.
- Examples
- Uses a 3 × 3 MNIST convolution and filter-activation optimization on a pretrained Xception model.
- What you can do after
- Diagnose/debug with it by inspecting a filter’s response map or optimizing an input image to reveal its preferred pattern.
Read how Deep Learning with Python, Third Edition explains convolution kernel →
Deep Learning with R, Third Edition
A convolution kernel is a learned weight matrix that transforms each extracted input patch into an output feature vector.
- How it is used
- Explains how convolutional layers extract local visual patterns and build feature maps for image classification.
- How deep it goes
- Explains patches, channels, response maps, padding, and strides, but stops short of formal weight calculations.
- Examples
- Uses MNIST dimensions, 3 × 3 and 5 × 5 windows, and channel visualizations of a cat image.
- What you can do after
- Evaluate learned patterns by inspecting convolutional activations and their spatial response maps.
Read how Deep Learning with R, Third Edition explains convolution kernel →
Graph Neural Networks in Action
A learnable weighted operation that transforms input data to emphasize particular patterns or features.
- How it is used
- Uses learned filters to extract localized patterns for image tasks and graph-based node prediction.
- How deep it goes
- Explains the layer pipeline and spatial-versus-spectral distinction without deriving the convolution operator mathematically.
- Examples
- Uses an animal image, a pixel grid, graph nodes, and graph signals to contrast convolution settings.
- What you can do after
- Understand how learned filters transform grid or graph data into task-relevant representations.
Read how Graph Neural Networks in Action explains convolution kernel →
Learn Generative AI with PyTorch
A learned convolutional kernel whose weights slide across local input regions to produce feature responses.
- How it is used
- Uses learned kernels for spatial feature extraction in convolutional neural networks, especially image classification and generation.
- How deep it goes
- Explains sliding multiplication, shared weights, tensor shapes, layer arguments, parameter replacement, and numerical output, but not kernel training dynamics.
- Examples
- Demonstrates a 2 × 2 kernel scanning a 3 × 3 grayscale tensor and producing a 2 × 2 output.
- What you can do after
- Build it from scratch in PyTorch by defining Conv2d arguments, assigning kernel weights, and checking the resulting feature map.
Read how Learn Generative AI with PyTorch explains convolution kernel →
Sutskever's List
A convolutional kernel is a grid of learned weights applied across local regions of an input to produce spatial responses.
- How it is used
- Uses learned local patterns to extract increasingly complex visual features in AlexNet’s convolutional layers.
- How deep it goes
- Explains sliding dot-product comparisons, feature maps, and shared weights, but does not detail how the weights are learned.
- Examples
- Illustrates edge and cat-part responses, plus AlexNet’s progression from 11 × 11 to 3 × 3 filters.
- What you can do after
- Understand how shared filters produce spatial feature maps and support hierarchical visual patterns.
Evolutionary Deep Learning: Genetic algorithms and neural networks
A convolutional filter is a learned array of weights that is slid across local input regions to produce feature responses.
- How it is used
- Uses learned local filters to extract image features within convolutional neural networks.
- How deep it goes
- Explains sliding, stride, learned weights, channel changes, and pooling, but avoids mathematical derivation.
- Examples
- Demonstrates Fashion-MNIST image patches, a visual filter operation, and a three-layer Keras Conv2D stack.
- What you can do after
- Build convolutional image layers with Keras Conv2D and inspect their output shapes and parameter counts.
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.
- convolutional layerused by · A convolutional layer is a neural-network layer that applies shared learned filters to local regions of structured input to produce feature representations.
- 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.
- image processinguses · Image processing is the computational manipulation and analysis of digital image data to transform images or extract information.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.