Definition icon
Definition

Bottleneck Residual Blocks

A bottleneck residual block replaces the standard two-layer residual function with three convolutions: 1×11 \times 1, 3×33 \times 3, and 1×11 \times 1. The first 1×11 \times 1 convolution reduces the channel dimension, the 3×33 \times 3 convolution processes this lower-dimensional representation, and the final 1×11 \times 1 convolution restores the channel dimension. When the block's input and output dimensions match, a parameter-free identity shortcut connects its high-dimensional endpoints; where dimensions change in these bottleneck ResNets, a projection shortcut performs dimension matching. This bottleneck structure enables deeper ResNet configurations while controlling computational cost.

0

1

Definition icon
Updated 2026-09-19

Tags

Prep Sessions

Foundational Deep Learning Architectures: Transformers and Residual Networks @ University of Michigan - Ann Arbor

Ch.3 Deep Residual Network Architecture - Foundational Deep Learning Architectures: Transformers and Residual Networks @ University of Michigan - Ann Arbor

Residual Formulation and Shortcut Connections - Foundational Deep Learning Architectures: Transformers and Residual Networks @ University of Michigan - Ann Arbor

Identity versus Projection Shortcuts - Foundational Deep Learning Architectures: Transformers and Residual Networks @ University of Michigan - Ann Arbor

Deeper Bottleneck Architectures - Foundational Deep Learning Architectures: Transformers and Residual Networks @ University of Michigan - Ann Arbor