Module 8: Deep Learning


Topic 2: Convolutional Networks

Convolutional networks/Deep learning

We are going to do a brief overview of convolutional neural networks and I will also provide several other overviews.  As I’ve mentioned, the deep dive into these come in the advanced machine learning class.  What I want you to learn here is the general idea so that you have a high-level understanding of how convolutional neural networks work.

There is one reading from the book but a bunch of extra readings that you could do if you are interested and want to learn more!

  • Book reading: Read Section 21.3 (Convolutional Networks)
    • Note, the rest of the chapter is excellent but we are just doing an overview here (since you cover these in more depth in both the Machine Learning class or the Advanced Machine Learning class).  
  • Some additional reading if you are really interested in deep learning
    • Neural Networks and Deep Learning – this is a fabulous online and free textbook resource!  I highly recommend it if you want to learn more about neural networks and about deep learning as an excellent starting point.
    • Deep Learning – this is also a fabulous book with a much deeper level of math required.  If you are excited by the topic and want to learn more with all the math, this is a good book
    • Deep Learning with Python – this is a practical hands-on book about doing deep learning with python.  There are free Jupyter notebooks online from the book as well to get started.  
    • This is an article that contains a lot of links to other deep learning resources A “weird” introduction to deep learning 

Introduction to convolutional networks

Copy of my slides

  •  

Another introduction to deep learning

The video below provides a nice high-level introduction to deep learning 

Additional resources

  • If you really love deep learning and want to learn more, the books are listed above and you can find in-depth online courses here:
    • Coursera course on deep learning  
    • deeplearning.ai have more than 6 courses on deep learning, convolutional networks, and advanced topics
    • While this YouTube channel isn’t all deep learning, a large number of the papers that he covers are deep learning related.  Although it is called two minute papers, each is about 5 minutes.  Very fun and informative! Two minute papers

Exercise

Go to the quiz on CNNs