Computer Vision | Transfer Learning and Pre-Trained Models | Lecture 13

Transfer learning is one of the most exciting paradigms in modern machine learning - The ability to re-use learning from one context to other and not have to re-invent the wheel with big DNN/CNN architectures.

In this lecture, we discuss strategies for transfer learning and using pre-trained models in the context of Convolutional Neural Networks.

0:00 Introduction and Logistics
5:16 Transfer Learning for CNNs
23:15 3 Models for Transfer Learning
26:56 Why Transfer Learning?
28:52 CNN Code
31:39 Nuanced Strategies
37:18 How to pick a strategy?
43:29 ICE #1
47:12 Pytorch Transfer Learning Notebook
1:12:28 ICE #2