Stride & Padding

What is Stride & Padding?

Stride is the step size of the filter. Padding adds border pixels (usually zeros) to control the output size. Together they determine the dimensions of the feature map.

Where did the term "Stride & Padding" come from?

Hyperparameters for convolution algebra.

How is "Stride & Padding" used today?

Critical for designing deep architectures like ResNet.

Related Terms

• Convolutional Neural Network (CNN)
• Kernel (Filter)