fa25-cs540 cs540outputinput
Lecture 15: Deep Learning I - Convolutional Neural Networks Lecture 15: Deep Learning I - Convolutional Neural Networks
Content locked
The catalog is public. Reading this content requires approved access.
Sign in to readLecture 15: Deep Learning I - Convolutional Neural Networks
Review: Multi-Layer Neural Networks
Motivation: The Problem with Fully Connected Networks
Why Convolution?
2-D Convolution
Basic Operation
2-D Convolution Layer
Example Kernels
Convolutional Neural Networks
Advantage: Sparse Interaction
Padding
The Shrinking Problem
Solution: Padding
Padding Formula
Stride
Stride Formula
Multiple Input and Output Channels
Multiple Input Channels
Multiple Output Channels
Pooling Layer
Max Pooling
Average Pooling
Pooling Properties
Practice Problems
Key Takeaways
Suggested Reading