Strided

• 
• Normally S = 1
• S>1 → Downsampling
• Dilated
• Spaces in the filter kernel
• D = 1 : normal Conv aka D-1 spaces
• Effective Filter size : $\hat{F}=F+(F-1)(D-1)$
•