A convolution performed with stride greater than one so it reduces spatial resolution instead of using a separate pooling layer.
Strided convolution reduces an image’s spatial resolution inside a convolutional layer, providing an alternative to a separate pooling layer. section 8.6
Apply it by:
Replacing pooling operations in a U-Net with strided convolutions to create a fully convolutional model. section 8.6
Comparing standard and anti-aliased strided convolutions with max pooling and average pooling on CIFAR-10. section 14.4
Architecture experiment
Modify a U-Net by replacing all pooling layers with strided convolutions, then compare its performance with the original design. section 8.6
How Inside Deep Learning: Math, Algorithms, Models covers it
How deep it goes
Practices architectural replacement and anti-aliased implementation, but does not derive convolution output shapes or internals.
Examples
Uses U-Net replacement experiments and CIFAR-10 comparisons involving standard and anti-aliased variants.
What you can do after
Build it from scratch by replacing pooling with standard or anti-aliased strided convolutions and evaluating performance.
How it connects
Connects strided convolution with max pooling, average pooling, U-Net, residual networks, and anti-aliasing.
Deep Learning with R, Third EditionA strided convolution is a convolution that moves its kernel by more than one input location at a time to reduce spatial resolution.
max poolingcontrasts with · Max pooling is an aggregation operation that replaces each local or grouped region with its maximum value, often to downsample a representation.
convolutionkind of · Convolution is an operation that applies a kernel across an input, combining values from overlapping local regions with shared weights to produce outputs.
downsamplingimplements · Downsampling is the reduction of data quantity, spatial resolution, temporal resolution, or sampling rate by retaining, selecting, or aggregating fewer representative values or observations.
strideuses · A stride is the step interval between successive positions or elements accessed, selected, or processed along a dimension or sequence.
This explainer was generated by Manning's knowledge graph from the text of Inside Deep Learning: Math, Algorithms, Models. The section links point to the parts of the book it draws on.
