Module 5 – Advanced CNN Architectures

Name: Lecture 5 – Modern CNN Architectures and Transfer Learning
Description: Lecture video and resources for Module 5 of the Deep Learning course by DYNAMO Lab.

What you'll learn

Builds on Modules 1–4

≈6 h · 2 h video · 2 h reading · 2 h coding

Inspect gradient flow; compare networks with/without BatchNorm
Implement a ResNet block; reproduce a mini-ResNet training
Fine-tune a pretrained model; summarize efficiency trade-offs (e.g., MobileNet)

Lecture 5 – Modern CNN Architectures and Transfer Learning

📚 Resources & Lecture Code

Required reading · course book

Ch. 9 · Modern CNNs and Transfer Learning
Appendix B · Tensors in Practice
Appendix C · Numerical Precision and Hardware Efficiency
Appendix D · Notation

The Colab notebook contains the lecture code for Module 5 (advanced CNN architectures and transfer learning). Use it to follow along the in-class demos.

Open Colab Notebook (shared for Modules 4–5)

Module 4

Module 6