🔢 CNN Operations Deep Dive
94 words
1 min read
Visual companion
Python
Type and operator map
Python Week 1: the first filter for runtime behavior
View
Revision summary
What this note is really saying
Short form
# 🔢 CNN Operations Deep Dive ## 1. 🎯 Learning Objectives - Compute convolution output dimensions - Count parameters in CNN layers - Implement convolution in Python ## 2.
🔢 CNN Operations Deep Dive
1. 🎯 Learning Objectives
- Compute convolution output dimensions
- Count parameters in CNN layers
- Implement convolution in Python
2. 📖 Core Content
3.1 Convolution Output Size
H_out = ⌊(H_in + 2P - F)/S + 1⌋
3.2 Parameter Count
Conv layer: (k_h × k_w × C_in + 1) × C_out FC layer: (n_in + 1) × n_out
3.3 Example: AlexNet
Conv1: 11×11×3, 96 filters, stride 4, pad 0 Input: 227×227×3. Output: (227-11)/4+1 = 55. 55×55×96 Parameters: (11×11×3+1)×96 = 34,944
Join Discord
PreviousRegularizationNextCNNs