Feature map
A feature map is a spatially organized array of neural-network activations representing the presence or strength of learned features at different locations.
Deep Learning with R, Third Edition
A feature map is a spatial array of activations indicating the presence or strength of a learned visual pattern at different image locations.
- How it is used
- Carries location-specific filter responses between convolutional layers and into image classification, feature extraction, segmentation, and object detection.
- How deep it goes
- Defines tensor axes and per-channel responses, then follows shape changes, pooling, reuse, and visualization without a mathematical derivation.
- Examples
- Traces MNIST feature-map shapes and plots a channel activated by a cat image.
- What you can do after
- Apply it via R/Keras to read convolutional output shapes and inspect individual activation channels.
Read how Deep Learning with R, Third Edition explains feature map →
Grokking Machine Learning
A mapping that sends data points from an original input space into another feature space, usually of higher dimension.
- How it is used
- Uses higher-dimensional representations to make nonlinear datasets linearly separable for SVM classification.
- How deep it goes
- Explains geometric mapping and polynomial and RBF examples, while deferring the full theory of kernels and feature maps.
- Examples
- Uses x1x2, quadratic columns, similarity columns, XOR-style points, circular datasets, and projected planes or curves.
- What you can do after
- Apply it via SVC with polynomial or RBF kernels to classify datasets with curved boundaries.
Grokking AI Algorithms, Second Edition
A feature map is a spatially organized array of activation values representing learned features at different locations in an input or intermediate image representation.
- How it is used
- Carries location-aware filter detections through convolutional layers and between U-Net encoder and decoder stages.
- How deep it goes
- Works through sliding-filter dot products, padding, biases, and an eight-map tensor, then follows maps through downsampling and upsampling.
- Examples
- Traces eight 3×3 filters over an 8×8 noisy image and illustrates high-resolution maps joining upsampled maps.
- What you can do after
- Build it from scratch in NumPy by padding an image and sliding filters to calculate feature-map values.
Read how Grokking AI Algorithms, Second Edition explains feature map →
Math and Architectures of Deep Learning
A spatial array of activation values produced by applying a convolution or pooling operation over an input image or prior layer output.
- How it is used
- Carries spatial convolutional representations into pooling, classification, and region-based object detection.
- How deep it goes
- Explains spatial summaries, pooling resolution changes, receptive fields, and RoI extraction without deriving convolution arithmetic.
- Examples
- Uses LeNet map dimensions, 2 × 2 max and average pooling, stacked 3 × 3 convolutions, and RoI pooling.
- What you can do after
- Evaluate/understand how convolutional representations retain spatial information while supporting classification and region feature extraction.
Read how Math and Architectures of Deep Learning explains feature map →
Deep Learning with Python, Third Edition
A spatial grid of activations produced by a convolutional or pooling layer, where each channel encodes the presence of a learned feature over locations.
- How it is used
- Tracks filter responses across image locations as convolutional and pooling layers build representations for image classification.
- How deep it goes
- Defines rank-3 shape and channel meaning, then works through convolution, pooling, and activation shapes without deriving filter calculations mathematically.
- Examples
- Uses MNIST digit shapes, a small ConvNet’s layer summary, and separately plotted channels from a cat-image activation.
- What you can do after
- Apply it via Keras to read ConvNet output shapes and visualize individual channels of intermediate activations.
Read how Deep Learning with Python, Third Edition explains feature map →
Sutskever's List
A spatial array of responses produced by applying a convolutional filter to an input.
- How it is used
- Represents where convolutional filters detect visual patterns in an image.
- How deep it goes
- Defines feature maps through filter responses and traces their progression from edges to object parts.
- Examples
- Illustrates 96 first-layer AlexNet maps and the progression from edge responses to contours and object parts.
- What you can do after
- Understand how convolutional filters produce spatial responses that deeper layers can combine into more abstract visual patterns.
Learn Generative AI with PyTorch
An intermediate activation map in a neural network whose spatial resolution can be increased or preserved by convolutional operations.
- How it is used
- Uses convolution outputs to represent detected spatial patterns in image-processing and image-classification pipelines.
- How deep it goes
- Explains filter scanning, arithmetic, tensor shapes, and PyTorch parameters, stopping short of deeper network architectures.
- Examples
- Demonstrates a 3×3 grayscale tensor, a 2×2 filter, fixed weights, and a resulting 2×2 output.
- What you can do after
- Builds a small feature map in PyTorch and verifies its values against manual convolution arithmetic.
Read how Learn Generative AI with PyTorch explains feature map →
Machine Learning Algorithms in Depth
A feature map is a function that represents input objects as vectors in a feature space, possibly an infinite-dimensional one.
- How it is used
- Represents inputs for learned convolutional responses and kernel-based inference.
- How deep it goes
- Derives convolutional construction and pooling roles, but leaves infinite-dimensional feature maps implicit rather than explicitly representing them.
- Examples
- Uses sliding-kernel outputs, LeNet on MNIST, max pooling, and Gaussian-kernel feature spaces as concrete examples.
- What you can do after
- Evaluate how convolution and pooling transform inputs, and understand when kernel evaluations replace explicit feature vectors.
Read how Machine Learning Algorithms in Depth explains feature map →
Related concepts
- convolutional layerproduced by · A convolutional layer is a neural-network layer that applies shared learned filters to local regions of structured input to produce feature representations.
- kernel functionproduced by · A kernel function is a mathematical function that measures similarity between pairs of inputs, often representing an inner product in an implicit feature space.
- convolution kernelproduced by · 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.
- Region Proposal Networkused by · A Region Proposal Network is a neural-network module that generates candidate object regions from image features for a two-stage detector.
- channelhas part · A channel is an independent component or dimension of image or tensor data carrying a distinct color, feature, activation, or value stream.
- max poolingused by · Max pooling is an aggregation operation that replaces each local or grouped region with its maximum value, often to downsample a representation.
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.